Connected Auto - FALL 2016 by Connected Auto

FALL 2016

PREMIER ISSUE

Designing

Connected Vehicles with

NXP Secure Interface Solutions Traffic Signals and Autonomous Vehicles

Utilizing Logic in Automotive Applications

Moving Autonomous Cars from the Lab to the Road

Impact of Automated Vehicle Technologies on Driver Skills â&#x20AC;&#x201D;and more

Connected Auto

ACCELERATING

INNOVATION Wind River and Intel drive the future of transportation

Everyone knows the automotive landscape is changing. Less clear is how to make the most of that change. Wind River is helping build the highway to our transportation future. Powered by deep expertise in industries demanding rigorous levels of safety and security, Wind River and Intel are actively working on automotive technologies that speed the development for tomorrow’s connected and autonomous cars.

www.windriver.com/auto

Fall 2016

FUTURAMA Almost 80 years ago, General Motors unveiled its vision of the automated Highway of the Future in its “Futurama” exhibit at the 1939 New York World’s Fair. Half a century later, the California Partners for Advanced Transportation Technology (PATH) was founded to make the vision of automated highways and intelligent transportation systems a reality. Today, PATH, the Institute of Transportation, Berkeley (ITS), California Department of Transportation (Caltrans) and the Department of Motor Vehicles (DMV) have prepared the California highways and the infrastructure so that when the rubber hits the road—the road, and the consumer, are ready. This first issue offers a well-rounded selection of articles on the current state of the research and technology that’s going into the connected and autonomous vehicle throughout the United States. The article PATH at 30 celebrates three decades of defining intelligent transportation, while some of its programs, the California Vehicle Test Bed in Palo Alto and the Integrated Corridor Management project in Southern California demonstrate their commitment to creating safer and more efficient transportation systems. Wind River tackles the transition from research to reality in Moving Autonomous Vehicles from the Lab to the Road, and GoMentum Station near Silicon Valley is Redefining Mobility for transportation giants and tech startups on their 5,000 acre miniature city. Our in-depth research from the Michigan DOT and the Center for Automotive Research is a two-part research paper on the Impact of Automated Vehicle Technologies on Driver Skills. Traffic Signals and Autonomous Vehicles covers alternative approaches to providing traffic light data to autonomous vehicles, and parking is going to get a lot easier using Comtech’s Location Studio navigation in Seamless Location for Tomorrow’s Car. This issue’s Technology Partner is NXP, one of the world’s leading suppliers of products for the Automotive Industry. An interview with Michael Lyons on Utilizing Logic in Automotive Applications is followed by Designing Secure Interface Solutions for Connected Vehicles from the NXP portfolio of more than 700 products, including the NXP Logic Q100 Logic solutions previewed in Automotive Logic Parts Lead the Way. We hope this premier issue is informative and entertaining, and perhaps in these pages, you might even catch a glimpse of how far we’ve come to achieving our own “Futurama”. Glenn ImObersteg Thomas West Convergence Promotions PATH Co-Director

© 2016 Intel Corporation. All rights reserved. Intel and the Intel logo are trademarks of Intel Corporation in the U.S. and/or other countries. *Other names and brands may be claimed as the property of others.

ConnectedAuto.org

Connected Auto

Fall 2016

44 About Connected Auto: Convergence Promotions LLC and California PATH (Partners for Advanced Transportation Technology) at the University of California, Berkeley, have partnered to advance the technologies that connect vehicles to the surrounding infrastructure and other vehicles. The goal of the partnership is to provide synergy and communications between academia, public institutions, automobile manufacturers and technology companies in the connected and autonomous vehicle industry.

Advisory Committee:

Alexandre M. Bayen Director, Institute of Transportation Studies Thomas West Co-Director, California PATH

Connected Auto: Publisher: Glenn ImObersteg Convergence Promotions LLC glenn@convergencepromotions.com Design and Production AspenCore: Karissa Manske, Carol Smiley DR Design: Dave Ramos

All product names, descriptions, specifications, etc., are subject to change without notice. The University of California, California, ITS, PATH and Convergence Promotions take no responsibility for false or misleading information, errors or omissions. Any comments maybe addressed to the Publisher, Glenn ImObersteg, at: glenn@convergencepromotions.com.

ConnectedAuto.org

18 24 30

Sales: Ruby Brower rubyb@rtcgroup.com Copyrights and Credits: The Mashead, logo, designand format of Connected Auto are copyright 2016, Convergence Promotions LLC. The contents of this publication are the property of the companies whose articles appear within. No portion of this publication may br preproduced in whole or in part without the express permission of the publisher in writing.

4 10

Redefining Mobility Transportation Giants test autonomous vehicles on the real-world infrastructure of GoMentum Station.

Traffic Signals and Autonomous Vehicles Exploring a Vision-based or V2I approach to the effect of traffic lights on autonomous vehicles.

42 44

Moving Autonomous Cars from the Lab to the Road WindRiver on the progress that has occurred and what obstcales still need to be overcome to move to the open road.

Utilizing Logic in Automotive Applications An exclusive interview with Michael Lyons, the Technical Marketing Manager for NXP’s BL Logic Division.

PATH at 30 UC Berkeley’s Partners for Advanced Transportation Technology celebrates three decades of defning intelligent Transportation.

Seamless Location for Tomorrow’s Connected Car The new Location Studio solves enhanced navigation, real-time traffic and other crucial automotive location services.

California Connected Vehicle Test Bed The nation’s first dedicated short-range communication test bed is open to test connected vehicle technology and applications.

Impact of Automated Vehicle Technologies on Driver Skills A special in-depth research paper by the Center for Automotive Research and the Michigan DOT on how autonomous vehicles will affect driving behavior. Part 1 of 2 Parts.

Using FRAM to Build Smart Airbag Auto Applications The key technological advantages of using FRAM non-volatile memory technology in airbags for increased safety and control.

Designing Secure Interface Solutions for Connected Vehicles An overview of some of NXP’s 700 automotive solutions solutions for the interface and system management sector.

Integrated Corridor Management Creating safer and more efficient transportation systems through connectivity. This article explores the effects of corridor management in Southern California’s I-210 Freeway.

ConnectedAuto.org

Connected Auto

Fall 2016

Transportation giants and tech startups test their autonomous vehicles on the real-world infrastructure of

GoMentum Station.

Redefining MOBILITY By the Contra Costa Transportation Authority

ConnectedAuto.org

Less than one hour from Silicon Valley and San Francisco, tucked into the rolling hills of Contra Costa County, is a 5,000-acre miniature city known as GoMentum Station. Located within the former Concord Naval Weapons Station and still guarded by the military, it would be understandable if those passing by the site didnâ&#x20AC;&#x2122;t realize that its bucolic hills and quiet roads are now the epicenter of transportation and infrastructure innovation and research in the Bay Area.

ConnectedAuto.org

Connected Auto

Fall 2016

Founded in 2014, GoMentum Station is the product of a creative, collaborative, publicprivate partnership between the Contra Costa Transportation Authority (CCTA), Stantec, and the City of Concord to reimagine and repurpose the decommissioned Concord Naval Weapons Station as a site for testing the next generation of autonomous and connected vehicles and smart infrastructure. In just two years, this partnership has established GoMentum Station as the leader for testing hands-off-wheel, feet-off-pedal capabilities; smart cars that communicate with each other and their surroundings; and driverless, shared autonomous vehicles poised to transform transportation as we know it. The efforts of CCTA staff to identify solutions to Contra Costa residents’ most challenging problems led to our agency’s laser focus on redefining mobility through thoughtful decisions about our county’s infrastructure. As more attention is paid to the ways in which autonomous and connected vehicles are poised to redefine the way we experience mobility—the safety, efficiency, and environmental impacts of travelling through the world—more attention is being focused on the role infrastructure plays in facilitating and furthering this transformative technology. CCTA is not a technology company, but the creation of GoMentum Station has given our tiny agency a prominent role in leading the next generation of mobility.

ConnectedAuto.org

Currently, GoMentum Station partners include Honda, pioneering self-driving cars for the consumer market; Otto, whose mission is to increase safety and efficiency of commercial freight trucks by retrofitting existing semis to drive autonomously; and EasyMile, who is poised to solve the first-andlast mile problem around the globe with their shared, autonomous, zero emission shuttles. More partners are waiting in the wings to be announced in the next six months. These transportation technology innovators come from around the world to test at GoMentum Station because there is simply nowhere else that offers real-world infrastructure elements in a private and secure testing ground. GoMentum Station features: • 20 miles of paved roadway. Our roadways offer a diverse sampling of pavement textures, from the recently refinished to the 40-year old pavement and striping found in much of America. • All the infrastructure of a typical city: buildings, parking lots, stop signs, sidewalks, curbs, gutters and striping. • More complex elements of transportation infrastructures, like bridges, undercrossings, railroad crossings, cattle guards, profile and elevation changes and curvilinear alignment.

• A large grid system of streets, affectionately known as “Bunker City.” In the future Bunker City will feature sections striped according to European and Asian standards, allowing multinational partners to fulfill diverse testing needs at a single location. • A seven-mile spine road, with long straightaways that can be used to test at high speeds. • Twin-bore 1,400-foot long “true tunnels.” GoMentum Station is the only test bed in the country to feature “true tunnels” that are more than 14 feet in diameter. Unlike above ground tunnels, these two “true tunnels” cut through the hillside enabling testing of GPS reception. • A mild California climate that means testing can—and does—continue uninterrupted year-round. Any or all of these elements can be used to conduct the type of isolated-variable research that is virtually impossible in real-life traffic situations. Despite its popularity, GoMentum Station’s partnership program was designed to be selective. CCTA carefully chooses partners with projects and programs that align with the values and priorities of the agency and Contra Costa residents. These values are reflected in the four tenets of GoMentum Station:

1. Enhance safety: connected and autonomous vehicle technology— features like automatic braking and lane adjustment—are saving lives. At GoMentum Station we want to enhance access to those life-saving features and usher in the next generation of safer vehicles. 2. Make mobility more efficient: finding the best ways to move people and goods around the county. As the population grows, efficiency becomes more and more important. 3. Support a thriving economy: clean, safe, efficient, and affordable transportation is key to a thriving economy, whether it is moving goods or cutting down on commute times for workers.

CCTA carefully chooses partners with projects and programs that align with the values and priorities of the agency and Contra Costa residents.

4. Protect the environment: we believe that the electric motor is the way of the future. GoMentum Station wants to support the expanded use of zero emission vehicles by helping to develop a smart infrastructure —things like flash charging, inductive charging for vehicles, and more.

ConnectedAuto.org

Connected Auto

Imagine a community that overlays a subscriptionbased autonomous transportation service over a connected city.

In addition to putting Contra Costa on the map for autonomous and connected vehicles, GoMentum Station is a way for our agency’s commissioners and staff to keep their finger on the pulse of innovative transportation research. CCTA firmly believes in looking beyond transportation solutions that attempt to build our way out of congestion. Instead, we’re evaluating and developing our long-range plans to incorporate current and future technology applications. CCTA’s vision for the future centers on the use of emerging technologies and public-private partnerships to meet current and future transportation demands and reduce greenhouse gas emissions in Contra Costa County. To make that happen, we need to be on the cutting edge of transportation technology. Imagine a community that overlays a subscription-based autonomous transportation service over a connected city. A steady stream of detailed information about travel patterns, traffic, accidents, and more is collected in real time by this digital city. That information leads to more informed budget priorities and planning regarding, for example, what road needs to be repaired first or what culvert needs to be cleaned. If data is transmitted that six cars on Main Street have received a sudden jolt to their shock absorbers at the same spot, someone from the Public Works Department could be dispatched to fill a pothole before it grows larger or causes an accident—but only if public agencies take a forward thinking position on the increasingly reciprocal relationship between transportation infrastructure and technology—and continue to incorporate new innovations as they emerge. Through GoMentum Station, our agency gains a knowledge of what is new and what is next in the world of transportation technology, and we can make informed policy choices and maximize improvements to the ways residents are able to move around Contra Costa. For CCTA, GoMentum Station means we’re staying informed, making the best possible policy decisions, and maximizing taxpayer dollars. It has been exciting to see the Federal Department of Transportation prioritize

ConnectedAuto.org

forward-thinking policy measures, as evidenced by the release of a Federal Automated Vehicles Policy in September. Secretary of Transportation Anthony Foxx noted that “as the digital era increasingly reaches deeper into transportation, our task at the U.S. Department of Transportation is not only to keep pace, but to ensure public safety while establishing a strong foundation such that the rules of the road can be known, understood, and responded to by industry and the public.” President Obama recently editorialized about the potential for autonomous vehicles to save tens of thousands of lives a year and add to the quality of life for everyone—if developed thoughtfully. It is an exciting time for public agencies, like CCTA, to participate in and contribute to the conversation about how this new technology can best be put to use. Whether it’s an 81-year old grandmother who no longer drives but still wants to visit with her granddaughter or a solution for the first-and-last-miles challenges of public transportation, by staying abreast of new and emerging technology, public agencies like CCTA can allow for a reimagining of how we get where we need to go. The next time you’re in the rolling hills of Contra Costa County, know that the quiet beauty conceals the innovative work happening to test technology that it is going to have a transformative effect on our transportation systems, our cities, and our lives.

Michigan is delivering the future of transportation mobility today. Planet M is home to more vision, talent, research resources, and innovative collaboration than anywhere on Earth. Learn more by visiting us at www.planetm.com. #MichiganPlanetM

Connected Auto

Fall 2016

Traffic Signals and Autonomous Vehicles:

Vision-based or a V2I Approach? Drivers need to understand traffic lights, and this is just as true for autonomous drivers as it is for human ones. While there have been some optimistic predictions that traffic lights would vanish when autonomous driving becomes a reality1,2,3, this seems a bit farfetched unless the pedestrians become automated as well. There will also obviously be an extended transition period as digital counterparts gradually replace human drivers.

By Matthew L. Ginsberg

Connected Signals, Inc. 355 Goodpasture Island Road, Suite 200, Eugene, OR 97401-2119 ginsberg@connectedsignals.com

ConnectedAuto.org

Connected Auto

Fall 2016

If autonomous vehicles are to share traffic lights as a fact of life with the rest of us, how are they to identify where the lights are, and whether any particular light is red, yellow, or green? At a high level, there seem to be two approaches. In one, autonomous vehicles deal with lights by observing the same traffic lights as do the rest of us, a vision system of some sort. In the other approach, the information is made available to autonomous drivers using an entirely different data stream. Each of these approaches has at least two subtypes. A vision-based approach might use general vision, picking the lights out from the visual field and then determining their color. Alternatively, a vision system might use some more specialized technique, exploiting additional information to both reduce the complexity of the vision problem somehow and make the solutions more reliable as well. The “data stream” approach also has two subtypes: depending on whether existing data infrastructure is used (presumably the wireless Internet, as connected cars will surely be a reality long before autonomous cars are), or whether some new and at least partly dedicated mechanism is used instead (such as the long-promised V2I connectivity provided by Dedicated ShortRange Communication, or DSRC). In this article, we discuss each of these four possibilities. While it is clear that none of them has been fully realized at this point, many partial or related implementations exist, and we can draw reasonable conclusions from the successes and failures of that other work. Doing this is our goal in this article. Before proceeding, however, we should note that it is not our intention here to predict the future, to say, “Autonomous vehicles will interact with traffic lights in the following way.” In actuality, none of the four approaches seems likely to work in isolation and, at least initially, the makers of autonomous vehicles will need to either restrict their domains of operation or rely on a hybrid approach. Our hope is that by presenting the prospective

ConnectedAuto.org

strengths and weaknesses of each option, we can help to guide the design choices that makers of autonomous vehicles are facing.

Vision-based Approaches Identifying traffic-signal state using a type of vision system certainly appears to be the most natural approach. Unfortunately, computer vision is in general far more challenging than one would expect. The reason, roughly speaking, is that we (humans) use a tremendous amount of world knowledge in analyzing any particular visual image. As the picture from Denver below indicates, it’s hardly a matter of simply identifying red, yellow, or green circles in the field of view. To make matters worse, computer vision systems are most often installed in environments where their apparently inevitable (albeit perhaps infrequent) errors have no catastrophic consequences. Traffic lights are not like this: an autonomous vehicle needs to get all of the lights right all of the time. Figure 1. Visual features resemble lights Using vision to find traffic lights involves solving two separate subproblems. First, the lights must be found in the image. Second, the states of the lights need to be determined based on the information in the image.

Signal Location This is hardly the place to present an introductory course in vision processing, but at least some understanding of the techniques involved is needed if the difficulties are to make sense. Objects whose structures are known are typically found in images by first creating a “Canny” image, which is basically a blackand-white drawing of the edges of objects in the original frame4,5. These edges can be found because the color in the image typically changes abruptly from, for example, a yellow balloon to an off-white sky.

Having found the edges, one can look for objects of any particular shape (i.e., the circular shape of a traffic light) by looking for circles in the Canny image. A variety of techniques are available for doing this, the most common of which is known as the Hough transform6,7. While this all works well in theory, practice is generally not so benign. A traffic light, for example, may not be circular because its edge is obscured (perhaps by the sunlight deflector at the top of the light itself), or because perspective causes the shape to be somewhat elliptic. Computer vision methods are extremely sensitive to small changes such as these, and any of them can cause some particular signal to be overlooked entirely. Consider, for example, the image below, where even to a human it is not obvious at a glance which are the lights and which are not.

Using vision to find traffic lights involves solving two separate sub-problems. First, the lights must be found in the image. Second, the states of the lights need to be determined based on the information in the image. required to recognize objects that have been designed to be as globally uniform as possible. It is difficult to understand the strengths and weaknesses of any particular approach because the academic literature is quite sparse. Most players in the autonomous car space (automakers, Google, etc.) are incredibly tightlipped about the details of their techniques. John et.al11 does discuss the use of a deep learning system for these purposes, while the more recent survey article by Jensen et.al12 echoes our conclusion that these two competing approaches cannot yet be thoroughly compared. It may also be helpful to examine a non-traffic system solving similar problems and about which a bit more is known. The ArcAngel system developed by Sportstech LLC13 tracks basketballs in high-definition video in realtime in order to determine whether or not a particular shot is likely to go into the hoop.

Figure 2. Which are the lights, which are the cars? If the structure is not known, techniques from machine learning are generally used instead. The state of the art in computer vision is to use deep learning with convolutional neural networks8 to extract the features of any particular image; the image itself is then classified and analyzed hierarchically as in ImageNet9. This approach currently has the best classification rates of any system (including humans) on traffic sign recognition10. While Google, Baidu, Audi, Mercedes and Volvo have all announced the use of deep learning in their autonomous vehicle projects, the technique is computationally more intensive than the Canny/Hough combination and it is not clear that deep learning is

The goal here is very similar. Basketballs are round. They are a distinctive color. They are often partially obscured by the players. The information needs to be processed quickly. What Sportstech has found is that a combination of circle detection and color pruning produces the best results, and these lessons may well apply to traffic signals as well. Sportstech has also learned that it is impractical to search entire images in real time, and ArcAngel works by searching entire images as rapidly as possible. Between the frames that are analyzed in their entirety, the system looks for objects (be they basketballs or lights) that are nearby to objects that were successfully found in the last fully analyzed frame. These “complete” and “partial” analyses are interleaved on a multicore processor to ensure that a timely data stream is constantly available to the user.

ConnectedAuto.org

Connected Auto

Fall 2016

For traffic lights, there is an alternative possibility, which is to use information from existing data sources to place the light accurately in 3-dimensional space. That 3-dimensional placement can then be combined with GPS data to determine the lightâ&#x20AC;&#x2122;s expected position in the image. Having identified the expected location of the signal, the light can either be assumed to be at its specified location or (likely better, given the Sportstech results) the light can be assumed simply to be in a relatively small neighborhood of that location. That lights can, in fact, move somewhat during somewhat normal conditions is made clear in the image in Figure 3. (And note as well the extent to which the light itself no longer presents as a circle in this shot.) While Google (and presumably others) are taking this approach14,15, there are some circumstances (Figure 4) that simply cannot be addressed in this fashion, as Gomes has noted16.

Signal Phase Having identified the locations of the signals, it remains to determine their colors. In general, this can be expected to be fairly straightforward. But it can, for example, be too hot or too cold (Figure 6). The bottom line is that signals break. When this happens, but only when it happens, an autonomous driver needs to realize that the malfunction is in the signal and not in the vision system, and to then respond appropriately cautiously. In Figure 6, none of the drivers at the intersection will be able to determine the state of the light in question. But in other cases, only the vehicle in question may be having problems.

Figure 6. Temperature extremes

Figure 3. Signals moving in the wind

Figure 4. A signal that will not appear in any database

For this reason, other players appear to be using general vision techniques to ensure that temporary signals are found as reliably as permanent ones. But they must then deal with images such as Figure 5.

Figure 5. Unanalyzable?

ConnectedAuto.org

Figure 7. Lights in an unknown phase The differences between Figures 6 and 7 are crucial. In Figure 6, every other driver or pedestrian will be aware of the problem and will be responding similarly cautiously. In Figure 7, however, other actors will not only be aware of the current phase of the light, but will probably have no idea that the vehicle confronted with the image shown is not.

V2I-based approaches Given the difficulties that will need to be addressed by a vision-based approach to this problem, it seems likely that an alternative data source should be found. Indeed, given the current enthusiasm for connecting vehicles to the outside world generally, perhaps this is a more obvious approach to take. Direct provision of information to vehicles will also allow those vehicles to ascertain the states of lights that are currently invisible because they are far away, around a curve or otherwise obscured, and to obtain additional useful information, such as the fact that a pedestrian has pushed a button and is waiting to cross. There appears to be two viable approaches to getting traffic light information to vehicles directly. The older one is to have the signals communicate directly with the vehicles, installing a DSRC radio at each light and a receiver on each vehicle. This approach has been in the works for approximately a decade and centers on a standard known as SPaT (Signal Phase and Timing)17. In the United States in 2014, there were approximately 200 signalized intersections for which SPaT data was available via DSRC18. An alternative approach, suggested more recently, is to provide traffic light data to vehicles using existing infrastructure. Many urban traffic lights currently connect to Traffic Management Centers, or TMCs, and provide those TMCs with real-time information regarding signal state, vehicle and pedestrian calls, and so on. The TMCs are in general connected to the Internet and can push this real-time data out to third parties. This authorâ&#x20AC;&#x2122;s company, Connected Signals, has recently gone even further, offering a free hardware device that listens to the network information flowing between the traffic lights and the TMCs and then pushes out traffic signal-related data directly. Connected Signals currently gets realtime signal information from approximately 10,000 signalized intersections worldwide. When compared to vision-based approaches, this sort of direct data acquisition has both strengths and weaknesses. Most importantly, of course,

the current state of computer vision means that data acquired directly from the signals should simply be more accurate and more robust than data acquired via a vision system. But that may be a mixed blessing, because it is possible for a signal to fail on the street and the TMC to be completely unaware of the problem. Faced with the situation of Figure 6, for example, drivers on the street will realize that the signals have failed. Drivers getting their data automatically may not. There are also important differences between the DSRC and Internet-based approaches: 1. Coverage. While any signal or vehicle can be equipped with a DSRC radio, only signals that are already connected to TMCs can be expected to provide their state over the Internet. Currently, only some 100,000 of the 300,000 signalized intersections in the United States19 are connected to TMCs; roughly speaking, it only makes sense to do so when the signals need to coordinate in some way with neighboring signals. The TMC handles synchronization among the signals on its network even though the clocks on individual signals may drift. For many rural lights, it seems unlikely that they will ever be connected to the Internet in this fashion. (The 100,000 number is based on the fact that approximately half of the 300,000 signalized intersections are in urban locations, and our experience with individual municipal agencies suggests that approximately 70% of urban lights are connected.)

An alternative approach, suggested more recently, is to provide traffic light data to vehicles using existing infrastructure.

2. Cost. While the primary argument against an Internet-based approach to this problem is coverage, the primary argument against the DSRC approach is simply cost. Estimates of the cost required to equip a single intersection range from $17,000 to $18,000 per intersection, plus backhaul costs of up to $40,000, and ongoing operations and maintenance costs of $2,000-3,000 annually18; a recent Econolite FAQ20 similarly suggested a cost of approximately $17,500 per intersection. While federal agencies have indicated a great interest in the DSRC approach generally,

ConnectedAuto.org

Connected Auto

Fall 2016

We expect the data-driven (as opposed to vision-based) approaches to use a hybrid of these two technologies. they have not shown a willingness to provide the approximately $5 billion in funding that would be required to equip all of the nation’s intersections with this functionality, and local or state agencies are less likely to have access to funds of this magnitude. 3. Protocol. One additional argument against the DSRC approach is that the SPaT protocol is relatively impoverished, including information only about the current phase of the signal in question. As autonomous drivers become more sophisticated, they will surely want access to additional information. Perhaps the simplest example is the one already mentioned involving pedestrians. If a pedestrian has pushed a button indicating a desire to cross in front of an autonomous vehicle, that vehicle should at least attempt to find that pedestrian in its field of view before proceeding through the intersection. This is true even if the autonomous vehicle has the right of way, since pedestrians are not always patient. These latter two points, together with a variety of additional technical concerns, led to a congressional peer review of the USDOT’s report on DSRC to take a relatively pessimistic view of the technology21.

All told, however, we do not view the DSRC and Internet-based approaches as competitive. For the 100,000 signalized intersections that are already connected, there is surely no reason to spend $1.75 billion installing DSRC radios. But for the 200,000 intersections that remain, DSRC may well turn out to be the method of choice. We expect the data-driven (as opposed to vision-based) approaches to use a hybrid of these two technologies.

Summary In this paper, we have described a variety of techniques for getting traffic-light information into autonomous vehicles. Two of those techniques have been vision-based (depending on whether the light locations were found visually or those locations were found in a database) and two have been data-driven (DSRC and Internet). We have also presented a variety of situations in which these techniques can be expected to fail, roughly corresponding to the figures that have appeared throughout the paper. If we summarize the abilities of the four methods to deal with the various failures, it looks something like this:

Vision Figure

ConnectedAuto.org

Description

Location visually

Data-Driven

Location from data

DSRC

Internet

Visual features resemble lights

Too many lights

Lights move

Temporary Lights

Lights broken

Lights can’t be seen

It should be obvious that no single method is a panacea, and a safe autonomous vehicle will almost inevitably need to use a combination of approaches. To the extent that specific conclusions can be drawn, we expect them to be the following: 1. Although the two data-driven approaches appear to be competitors, it is probably most appropriate to view them as likely providing similar functionality in different locations. We would expect the Internetbased approach to reach the most signalized intersections the most quickly, and the DSRCbased approach to provide supplementary coverage as the standard becomes more general and funds become available. 2. The “vision-based, location from data” approach strikes us as the least likely to be part of deployed systems, because this approach is incapable of dealing with many of the issues addressed by a purely vision-based approach, and also because this approach appears to be (mostly) dominated by the data-driven approach in locations where data is available. Of course, there is no particular reason to believe that the examples we have chosen to present will be representative, or that all of the problems will fit naturally into one of the groups in the above table. And here, perhaps, is the single greatest lesson that can be learned from the considerations we have presented: Autonomous driving is hard. It requires advancements in both engineering and in science. An intellectual environment in which all of the players guard progress so closely is unlikely to be in the interests of any of us.

Acknowledgement I would like to thank all of the members of the Connected Signals team for creating the stimulating environment in which we have jointly worked on these problems, and would especially like to thank David Etherington, Pamela Kinion and Tim Stirling for their help.

Bibliography 1.

Andy Boxall. “The traffic intersection of the future – busy, computer controlled and very nerve-wracking”, Digital Trends. [Online]; 2012 [cited 2016 2 29]. Available from: http:// www.digitaltrends.com/cars/the-trafficintersection-of-the-future-busy-computercontrolled-and-very-nerve-wracking.

13. Oliver Roeder. “Can An Astrophysicist Change The Way We Watch Sports?”, FiveThirtyEight. [Online]; 2015 [cited 2016 3 2]. Available from: http:// fivethirtyeight.com/features/ can-an-astrophysicist-changethe-way-we-watch-sports/.

Tim Beissmann. “Driverless cars could make stop signs and traffic lights disappear”, CarAdvice. [Online]; 2012 [cited 2016 2 29]. Available from: http://www.caradvice.com. au/159815/driverless-cars-could-makestop-signs-and-traffic-lights-disappear.

14. Nathaniel Fairfield, Chris Urmson. “Traffic Light Mapping and Detection”. In Proceedings IEEE International Conference on Robotics and Automation; 2011: IEEE. pp. 5421-5426.

Sarah M. Kaufman. “Farewell to traffic lights”, Satellite Magazine. [Online]; 2105 [cited 2016 2 29]. Available from: http://www.satellitemagazine. ca/2015/01/farewell-to-traffic-lights.

J. Canny. “A Computational Approach to Edge Detection”, IEEE Trans. Pattern Analysis and Machine Intelligence. 1986; 8(6): pp. 679-698.

The Wikimedia Foundation. “Canny edge detector”, Wikipedia, the free encyclopedia.[Online]; 2016 [cited 2016 3 2]. Available from: https://en.wikipedia. org/wiki/Canny_edge_detector.

R.O. Duda, P.E. Hart. “Use of the Hough Transform to Detect Lines and Curves in Pictures”, Comm. ACM. 1972 January; 15: pp. 11-15.

The Wikimedia Foundation. “Hough Transform”, Wikipedia, the free encyclopedia.[Online]; 2016 [cited 2016 3 2]. Available from: https://en.wikipedia. org/wiki/Hough_transform.

The Wikimedia Foundation. “Convolutional neural network”, Wikipedia, the free encyclopedia. [Online]; 2016 [cited 2016 3 3]. Available from: https://en.wikipedia. org/wiki/Convolutional_neural_network.

Jai Deng, Wei Dong, Richard Socher, Li-Jai Li, Kai Li, Li Fei-Fei. “ImageNet: A LargeScale Hierarchical Image Database”. In IEEE Conference on Computer Vision and Pattern Recognition 2009; 2009: IEEE. pp. 248-255.

10. Dan Cireşan, Ueli Meier, Jonathan Masci, Jürgen Schmidhuber. “Multicolumn deep neural network for traffic sign classification”, Neural Networks. 2012 Aug; 35: pp. 333-338. 11.

V. John, K. Yoneda, B. Qi, Z. Liu, S. Mita. “Traffic light recognition in varying illumination using deep learning and saliency map”. In IEEE 17th Conference on Intelligent Transportation Systems; 2014: IEEE. pp. 2286-2291.

12. Morten Borno Jensen, Mark Philip Philipsen, Andreas Mogelmose, Thomas Baltzer Moeslund, Mohan Manubhai Trivedi. “Vision for Looking at Traffic Lights: Issues, Survey, and Perspectives”, IEEE Transactions on Intelligent Transportation Systems. 2016; PP(99): pp. 1-16.

15. Sebastian Thrun, Chris Urmson. How Google’s Self-Driving Car Works. 2011 Oct 14. https://www.youtube.com/ watch?v=YXylqtEQ0tk at the 7:45 mark. 16. Lee Gomes. “Hidden Obstacles for Google’s Self-Driving Cars”, MIT Technology Review. 2014 Aug 28. 17. Bruce Abernethy, Scott Andrews, Gary Pruitt. Signal Phase and Timing (SPaT) Applications, Communications Requirements, Communications Technology Potential Solutions, Issues and Recommendations. Final Report. ARINC Incorporated; 2012. Report No.: FHWA-JPO-13-002. Available from:http://www.its.dot.gov/research/ pdf/spat_recommendations.pdf. 18. James Wright, J. Kyle Garrett, Christopher J. Hill, Gregory D. Krueger, Julie H. Evans, Scott Andrews, et al. National Connected Vehicle Field Infrastructure Footprint Analysis: Final Report. Washington, DC: American Association of State Highway and Transportation Officials and Transport Canada; 2014. Report No.: FHWAJPO-14-125. Available from:http:// stsmo.transportation.org/documents/ aashto%20final%20report%20_v1.1.pdf. 19. US Department of Transportation. Intelligent Transportation Systems for Traffic Signal Control: Deployment Benefits and Lessons Learned. 2007. Report No.: FHWA-JPO-07-004. Available from: http://ntl.bts.gov/ lib/jpodocs/brochure/14321_files/ a1019-tsc_digital_n3.pdf. 20. Econolite Group, Inc. “Connected Vehicle FAQ; Econolite Group’s Commitment to Innovation”, [Online]; 2015 [cited 2016 2 29]. Available from: http://www. econolite.com/files/1314/3447/1235/ Econolite-ConnectedVehicleFAQ.pdf. 21. Dennis Wilkie. Review of the Status of the Dedicated Short-Range Communications Technology and Applications [Draft] Report to Congress. Letter Report. Washington, DC: The National Academies, Transportation Research Board; 2015. Available from: http://onlinepubs.trb.org/onlinepubs/ reports/dsrc_april_28_2015.pdf.

ConnectedAuto.org

Connected Auto

Fall 2016

Moving Autonomous Cars from the to the

Lab

Road

What progress has occurred, and what obstacles remain in the way?

By Marques McCammon, General Manager, Connected Vehicle Solutions, Wind River;

Today, it’s not that helpful to postulate any particular timeframe,

and Lee Barnes, Director, Connected and Autonomous Vehicle Business, Ricardo

few—and partly because “autonomous car” means different things to

partly because there are too many variables—social, cultural, technological, economic, regulatory, and ethical, to name just a different people. What is far more helpful is to take a closer look at what autonomous cars are today, what they will become, and what the core requirements will be for moving them out of the lab and onto the open road. So let’s break it down into a few meaningful categories.

ConnectedAuto.org

Connected Auto

Fall 2016

Level 0: Where most cars are today. The driver has his/her eyes on the road and hands on the wheel, and is continuously exercising longitudinal and lateral control. Level 1: Some portion of longitudinal or lateral control is assisted by autonomous capabilities such as motion control or steering controllers. Level 2: The driver can occasionally take the hands off the wheel and allow autonomics to take control in a specific, limited use case, such as driving from home to the pharmacy. Level 3: The driver no longer has to monitor the system at all times but must be in a position to resume control, within a sufficient time margin, if required to handle an unexpected situation.

The point is that the autonomous car will evolve gradually over a long period. And it will evolve through the delivery and refinement of specific autonomic features and capabilities, which can also be categorized. These would include safety features such as forward collision warnings, advanced emergency braking, and pedestrian detection; driver monitoring such as fatigue detection; convenience features such as adaptive cruise control or traffic jam assistance; or vision-related features such as 360-degree visibility or night vision assistance. The big question—and the real source of magic for the delivery of autonomous capabilities— is how these features will be combined and how intelligence will be added to enable more and better automated functionality. And the answer to that question, in a word, is software. Software is driving the value chain of the autonomous car. Software will determine how the car plots its course, how it turns, when it changes lanes, even which music or videos are available to suit the consumer’s preferences. At least half of the value of the autonomous car will be derived from software in the near future. And that means automotive original equipment manufacturers (OEMs) will need to focus on developing a coherent strategy for software development, delivery, and validation. In other words, OEMs will begin to apply the same traditional V-shaped model they’ve used for system-level implementation to the new world of software engineering. In this modelbased development process, every step of the V model is iterated, from system requirement specifications to system design to component design to systems-vehicle integration.

ConnectedAuto.org

V&V of system behavior and nonfunctional system requirements

tio

Requirements Specification

Vechicle Integration

ula

Level 5: The system can cope with all situations automatically during the entire journey. No driver is required at all.

Functional Specification

System Integration Int eg r V& ation V /

First, the autonomous car is not a single, definable entity or set of specifications—it is a continuum of capabilities. There are multiple levels of automation. One way of describing the levels would be as follows:

Level 4: A driver is not required during a defined use case; the system can cope with all situations automatically in that defined use case.

Sim

Today, it’s not that helpful to postulate any particular timeframe, partly because there are too many variables—social, cultural, technological, economic, regulatory, and ethical, to name just a few—and partly because “autonomous car” means different things to different people. What is far more helpful is to take a closer look at what autonomous cars are today, what they will become, and what the core requirements will be for moving them out of the lab and onto the open road. So let’s break it down into a few meaningful categories. Autonomous Cars—Today and Tomorrow

System Design

Component Design Component Integration Development of Software & Hardware

Hitting the Accelerator on High-Quality Software Development and Validation

product specifications, concept development, and rapid prototyping, validation, and testing.

The V-shaped approach described above makes it possible to exchange information more rapidly within the overall vehicle and software development system. It also provides visibility to other members of the ecosystem, and it facilitates more interaction with suppliers. This, in turn, provides a mechanism for early validation of features and system architecture, and a continuous interchange of engineering artifacts.

Looking at validation and testing in particular, the move to agent-based modeling will be extremely helpful in accelerating the development of the autonomous car. Agentbased modeling is simply modeling as a collection of autonomous decision-making entities called agents. These agents can be pedestrians, vehicles, or infrastructure, for example. Each agent individually assesses its situation and makes decisions on the basis of a set of rules—for example, being self-directive. The agents are placed in an environment where a feature set—such as weather, infrastructure, roads, traffic signs, and so on—and the agents may execute behaviors that are appropriate for the system they represent. Finally, a simulation can be run, and out of this simulation comes an opportunity to see what happens when you place an autonomous vehicle in an environment where you have pedestrians, or severe weather conditions, or other variables, to determine how the system performs in that environment.

Clearly, automotive system engineering covers a wide variety of attributes, and there are many special considerations and processes that must be employed to ensure a robust and highquality product. The trick is to ensure that the emerging hardware and software systems can integrate smoothly with these vehicle systems, and this will require technical expertise and experience. The system engineering team must focus on the key activities to enable rapid product development, and these include requirements development to drive the key

ConnectedAuto.org

Connected Auto

This, in turn, allows engineers to test and validate both the component and the vehicle in those scenarios. Engineers can leverage something like Ricardo’s Agent Drive, which makes it possible to place that vehicle in an environment in which it is intended to operate and simulate different aspects of that environment to get the most accurate real-world scenario where the vehicles can operate and be tested.

• Legislation and regulation: Will the market have sufficient confidence in a software-based algorithm, and in artificial intelligence, to allow them to substitute for human judgment about the value of human life or physical assets? And will the auto industry—and its millions of jobs—need to be protected by legislation against the potential economic impact of the autonomous car, or will innovation outpace regulation?

Once the simulation is done, engineers can bring all of that learning to a common framework, reducing the need to have more hardware earlier in the process. Once the system has been validated in a neutral, cloudbased environment, it can be ported directly to the host hardware. In this way, a new kind of supply chain model becomes possible—one that’s consistent with the way an automaker might use the simulation of a vehicle drive system and then source chassis components or seating and door components but now applied to a software-based environment.

• Safety: Will the advent of the self-driving car drastically increase the overall safety of the vehicle driving experience? Studies show a potential cost savings of $300 billion per year due to lower accident rates and higher safety, but how realistic is that?

The bottom line: Engineering teams that have the ability to understand, manipulate, and implement these key process requirements and innovations will be the relevant players in this new and emerging technology environment.

More Bridges to Be Built Of course, the technological hurdles represent only one category of challenges for moving the autonomous car from the drawing board to the road. There are many other bridges that must be built and questions that must be answered. To name just a few:

Connecting Global Competence

• Consumer sentiment: How will consumers accept the autonomous car? A recent survey done by AAA shows that most drivers still have fear and distrust of autonomous driving—but not on a scale that’s unexpected given the magnitude of the change and its impact on everyday life.

electronica Automotive Conference.

• Recognition of the benefits: So far the focus has been on overcoming the negatives—but how about acknowledging the benefits of autonomous driving? At what point will society feel that the advantages of lower congestion, fewer accidents, higher productivity, efficiency gains through ride sharing, reduced emissions, increased accessibility, and so on will outweigh the short-term negatives?

International Conference on Technologies and Strategies for Automotive Electronics and Components. Topics: Safety and security Autonomous driving Interior electronics

• Business models: How will the autonomous car create new options such as “mobility as a service,” or the ability to select a different type of car each day, week, or month depending on the consumer’s mood or financial circumstances— and who will the winners be in this new era? Wind River® and its partners such as Ricardo do not claim to have all the answers to these questions, but we are excited to be at the forefront of exploration—driving the evolution of the software that will accelerate the adoption of autonomous vehicles.

Information & Registration: electronica.de/en/automotiveconference

List of speakers (excerpt):

Simon Fürst

BMW Group for AUTOSAR

Andreas Klage

DRÄXLMAIER Group

Dr. Ludger Laufenberg Kostal

Wolfgang Lenders BMW Car IT

Steve Nadig Daimler Trucks

electronica Automotive Conference November 7, 2016 I Messe München

ConnectedAuto.org

The conference is held within the scope of electronica, the World’s Leading Trade Fair for Electronic Components, Systems and Applications.

Dr. Reinhard Ploss Infineon Technologies

Dr. Stefan Poledna TTTech

Martin Schleicher Elektrobit

Dirk Wollschläger IBM

Connected Auto

Utilizing LOGIC in Automotive Applications An Exclusive Connected Auto Interview with NXP’s Michael Lyons, the Technical Marketing Manager for BL Logic

Fall 2016

LOGIC, that special form of digital circuitry that allows different

types of chips or circuits to work together by acting as an interface between them, has been around since the days when engineers used slide rules, but is still an essential part of embedded design because its inclusion generally occurs near the end of a project. Only after the key chipset decisions have been made does it become clear which logic functions are required to support small modifications and finetune performance, as once a chip’s design is finalized its functionality cannot be changed. The only way to implement last minute additions and improve time-to-market is to add this discrete circuitry; hence innovation in logic components is critical to advancing electronic design. In automotive electronics, logic supports applications by implementing simple motor control, smart key detection, security alarms, speed alerts, and more in an environment where temperatures could start at lower than -20 ˚C and, in some cases in the engine compartment, approach 150 ˚C. Given this range, to position parts for the best possible design success the Automotive Electronics Council developed the AEC-Q100 standard, which outlines procedures to be followed to ensure ICs meet the quality and reliability levels needed for automotive use. Applying quality standards that exceed AEC-Q100 for its logic portfolio, NXP has become the number one global supplier of logic for automotive applications. NXP’s logic products are used in a wide variety of automotive applications including instrument clusters, body control modules and engine control units. The company offers logic functions including (dear reader, take a breath here): analog switches, buffers/inverters, bus switches, counters, decoders/demultiplexers, multiplexers, flip-flops, gates, latches, level shifters, multi-vibrators, Schmitt-triggers, shift registers and transceivers. We recently spoke with Michael Lyons, Automotive Marketing Manager of NXP’s Logic business line, to find out how NXP intends to maintain its lead in developing, manufacturing, and qualifying automotive logic parts. Lyons has over 25 years of experience in new product development functions within the semiconductor industry.

ConnectedAuto.org

Connected Auto

Fall 2016

For the last 50 years, NXP’s logic business has supported the growing global demand for logic. How does the company ensure the best possible reliability and performance?

You supply a very large amount of commodity logic to much of the worldwide automotive industry. What keeps you ahead of the competition?

To improve reliability, we apply more conservative design rules. We do not use minimum-allowed dimensions and apply Six-Sigma design. Not designing to the limits of a process has proven to improve reliability. By using Six-Sigma design, any application the product is designed into is very robust; able to withstand a 1.5 Sigma shift in process. NXP has implemented a zero defect strategy to support automotive OEMs. This involves locating and removing outlier products. An outlier product is outside the normal distribution of a given production lot. As part of the zero defect philosophy outliers will not be sold as an automotive product even if within datasheet limits. Regular reliability monitoring is in place to ensure the process continues to produce quality and reliability

We have an automotive mindset and try to partner with customers. Automotive applications are long term applications that can be in production for many years. It is important to choose products at the beginning of a project that will still be in high volume at the end of the project. NXP advises automotive customers of what is driving trends in different product segments. This reduces the possibility of customers choosing low volume processes and packages. Staying with mainstream processes and packages maximizes quality and reliability.

Can you talk a bit about meeting the AEC-Q100 standard? We have a long history in automotive applications, first as Philips and now as NXP Semiconductors. We were a principal player in combining the various semiconductor qualification standards of automotive OEMs, into what is now known as AEC-Q100. This standard defines a minimum set of stress tests that must be performed successfully to qualify an integrated circuit product for automotive applications.

What about product/design support? What does NXP offer? Automotive is given priority product/design support. It is well known that the selection of control logic is often one of the last design decisions. We provide priority design-in assistance with Q100 qualification data available on request. If a customer has concerns there may be issues in their production released platform due to our device, we will review the case and provide a response within 24 hours. If the advice is to return the suspected device, inspection and any failure analysis activity will be given highest priority.

ConnectedAuto.org

As you just said, packaging matters, too. The lower gate count Mini Logic products reduce the footprint of many control logic applications.

Tell us a little about the Mini Logic option. The standard logic portfolio consists of devices that contain multiple functions. For example the 74LVC08APW is a quad 2-input AND gate in a TSSOP 14 package. In the past if a control logic application required a single AND gate there was no option but to use a device that contained four of them. The introduction of Mini Logic options has changed this. The Mini Logic packages house the same logic families as the larger SO, TSSOP & DQFN packages. Using only the number of gates required produces a significant saving in board space and simplifies board layout.

The discrete logic supplier must offer logic families that provide the required speed, performance at the lower voltages, have the lowest power dissipation capacitance (CPD) possible and have inputs that present the lowest possible capacitive load (CL). How have you met these challenges? Logic families are co-defined with customers. Newer logic families are often referred to as full feature families. This is because they include features that allow them to be used in partial power down (standby) applications. Due to the direct proportionality of power and voltage,

as power dissipation becomes more of a focus so does the ability to migrate applications to lower voltages. NXP logic has three full feature families that enable the migration of applications from 5.0 V down to as low as 0.8 V. LVC is fully specified from 5.5 V to 1.65 V. AUP is fully specified from 3.6 V to 1.1 V. Similarly, AXP is fully specified from 2.75 V to 0.75 V. AUP and AXP are specialized low power families. They are the lowest power logic families, facilitating the reduction of total system supply voltage in many applications.

Earlier this year at Embedded World 2016 in Nuremberg, Germany, you introduced the AXP family of logic translators, designed for lowpower and high-performance applications. What has been the market/customer response to these products? The voltage level translator products in our AXP logic family have received a lot of interest. We are seeing many applications for level translating gates in control logic. These translators allow significant reduction in main chipset supply voltages, while maintaining the ability to function with legacy peripherals. At the end of July we registered our first AXP automotive types.

In the run-up to the autonomous car there has been a lot of development in ADAS systems. A couple of examples of logic used in these applications come to mind: Automotive flipflops are used for timing synchronization to support virtual dashboard instrument clusters and video display boards. Adaptive headlamp design often requires discrete circuits, including multiple transistors, gate drivers and glue logic. Are their others? Yes, I’m also seeing voltage level translators and I/O expansion devices (multiplexers and shift registers) being used in these applications. I prefer to be less detailed to avoid damaging customer confidence.

NXP continually invests in new process and package technologies, as well as new packaging facilities, and is focused on increasing performance, lowering power consumption, and reducing size. The company has the largest portfolio of dedicated Q100 devices. As an example, consider its 74LVC08A-Q100 providing four 2-input AND gates. Inputs can be driven from either 3.3 V or 5 V devices allowing use of these devices as translators in mixed 3.3 V and 5 V applications. Features include: »» Automotive product qualification in accordance with AEC-Q100 (Grade 1) »» Specified from -40 °C to +85 °C and from -40 °C to +125 °C »» 5 V tolerant inputs for interfacing with 5 V logic »» Wide supply voltage range from 1.2 V to 3.6 V »» CMOS low power consumption »» Direct interface with TTL levels »» Complies with JEDEC standard: »» JESD8-7A (1.65 V to 1.95 V) »» JESD8-5A (2.3 V to 2.7 V) »» JESD8-C/JESD36 (2.7 V to 3.6 V) »» ESD protection: »» MIL-STD-883, method 3015 exceeds 2000 V »» HBM JESD22-A114F exceeds 2000 V »» MM JESD22-A115-A exceeds 200 V (C = 200 pF, R = 0 Ω) »» Multiple package options

ConnectedAuto.org

Connected Auto

The overall trend of integrating more and more functions on SoCs, does not seem to have had a negative effect on the market for your products. Would you agree with that assessment?

Nexperia will be the future name of the Standard Products business of NXP Semiconductors. Where are you in the transition and what impact (if any) will it have on your customers?

The trend to higher integration within an application and the aggressive timeto-market targets of that application combine to produce a cyclic nature to logic usage. In the continuing evolution of applications we may see logic completely integrated in one generation only to be used more widely for feature addition in the next generation. The sheer number of applications, all at various points of their evolution act to buffer the market from large peaks or troughs in logic demand. Due to the number of applications, logic demand remains relatively constant. It also presents us with integration opportunities, the AXP translators are an integration of translators and gates driven by footprintconfined application requirements.

Not sure I can say too much here, only that the Nexperia team is starting to take shape. Customer continuity is a very important focus. I don’t expect any serious impact to customer service and product quality levels between BU Standard Products and Nexperia.

Auto Solutions

Finally, the NXP logic business—starting as Philips Semiconductors—has supported the growing global demand for logic for over 50 years. What can we expect going forward? Nexperia, being a standard products company, will remain focused on logic. To learn more, visit the NXP automotive logic page at http://bit.ly/2dyOtss where you can download datasheets. You’ll also find helpful videos, links to training/events, and an order portal.

ABOUT MICHAEL LYONS Michael Lyons is currently the NXP Technical Marketing Manager for BL Logic. He has over 25 years of experience in new product development functions within the semiconductor industry. Michael has held various marketing, business development and product engineering management positions. He defined the world’s lowest power logic family: AXP.

Advanced eXtremely-low Power logic (AXP) provides a speed upgrade to the AUP logic family—and it does so without increasing dynamic power dissipation. Fully specified from 0.75 V to 2.75 V, AXP logic is suitable for existing 2.5 V and 1.8 V applications, while supporting the trend to the lower 1.2 V and 0.8 V voltage nodes. NXP’s Logic portfolio consists of the industry’s smallest, leadless DQFN package for gates, octals, and MSI functions simplifying PCB routing, and the world’s smallest MicroPak™ and GX packages for single-, dual-, and triple-gate logic. As a result designers can combine the logic function and package footprint that best suits the application.

Automotive reference application for Linux and QNX Location Studio delivers a robust platform for automotive products, providing extensive support for embedded navigation applications and the evolving 'Connected Car' environment. Location Studio provides a superior in-car navigation solution, which includes automatic map updates, real-time traffic, and integrated content such as real-time gas prices, Doppler weather radar overlay, movie theater locations, showtimes, ratings and reviews. It also captures analytics data that enables automotive manufacturers to improve the driver experience and develop next generation vehicles. With a carrier grade platform and over 10 years of connected navigation solutions experience, Location Studio has the technology and experience to provide a superior connected car solution today. © 2016 Comtech Telecommunications Corp. All Rights Reserved

ConnectedAuto.org

Learn more! Visit www.location.studio

Connected Auto

Fall 2016

PATH at

UC Berkeleyâ&#x20AC;&#x2122;s Partners for Advanced Transportation Technology celebrates Three Decades of defining Intelligent Transportation In the world of technology, a decade seems to signify a full generation, and looking ahead thirty years raises daunting uncertainties. In this context, it is sobering to recognize that the California PATH Program at the University of California Berkeley is celebrating its thirtieth birthday this year. These three decades represent the lifespan of intelligent transportation systems, which did not even have a name thirty years ago.

By Steven Shladover, Ching-Yao Chan, and Wei-Bin Hang

ConnectedAuto.org

Connected Auto

Fall 2016

In the mid-1980s, the California Department of Transportation (Caltrans) realized that they needed to look ahead a few decades and determine how they were going to be able to meet the continually growing transportation needs of a state with a rapidly growing population and economy. Traffic congestion was becoming an increasingly acute public concern, recognized as an impediment to the future economic health of the state. A planning study for the Los Angeles region considered a variety of alternatives for meeting the projected 20-year travel needs of the region. The only alternative considered in the study that could significantly ameliorate the growing congestion problems would have involved double-decking most of the major freeways, at the cost of $28 billion (in 1985 dollars). It was evident to all involved that this would be financially, politically, and environmentally infeasible, even before the seismic hazards of double-decker freeways were revealed by the 1989 Loma Prieta earthquake. This led to the important conclusion that “we can’t build our way out of congestion.” Very importantly, some of the people involved in the planning process recognized that there were possibilities for information technology to help with the problem, and decided to pursue that seriously.

This recognition among Caltrans and the leadership of the Institute of Transportation Studies at the University of California, Berkeley was the impetus for the creation of the PATH Program in 1986, beginning with exploratory studies of the possibilities and then expanding rapidly to become a full-scale research program with a diverse portfolio of research projects to address the many identified unknowns. Although the initial impetus for using information technology to improve road transportation operations came from California, the program founders at Caltrans and the University of California knew from the start that this was not something that California could do on its own. They knew that it would be necessary to create a national program to ensure nationwide interoperability of vehicles and to provide a large enough market for the new products and services that would be needed. Consequently, as soon as the program was created they devoted extensive efforts to missionary work in Washington DC and in other states and universities that had strong transportation research institutes and analogously strong relationships between their state departments of transportation and research universities. This led to a series of meetings among interested parties from across the U.S. during the period from 1987 to 1991, leading in turn to the creation of the Intelligent Vehicle-Highway Society of America (IVHS America, later renamed ITS America) and the inclusion of dedicated research funding in the 1991 surface transportation reauthorization bill, the Intermodal Surface Transportation Efficiency Act (ISTEA).

A new national program was needed to ensure nationwide interoperability of vehicles and provide a large enough market for the new products and services that would be needed. The research at PATH was initially organized in three major topic areas: • Navigation (which we now know as traffic management and traveler information) • Road vehicle automation • Roadway electrification (inductive power transfer from the roadway to battery electric vehicles in motion, enabling them to recharge their batteries). Although many other research programs around the country focused on traffic management and traveler information systems, the automation and electrification elements were unique to PATH. The electrification research lasted for only a few years, after several cycles of design and testing revealed that it was not feasible to install an efficient inductive power transfer system in the roadway at an affordable cost. Within the last few years with the advent of high-frequency power electronics and high-specific-power batteries, researchers in several countries have built on the earlier PATH research to develop a new generation of roadway electrification systems.

The PATH research in the other two topic areas has continued through the present, under a variety of different labels and internal management structures. Because of PATH’s pioneering status as the first ITS research program in the U.S., it has been able to attract the most talented faculty, graduate students, post-doctoral researchers and research staff to push the frontiers of state of the art. Rather than relying only on people from the traditional transportation disciplines of civil engineering and urban planning, PATH has attracted strong participation from highertech electrical engineering, computer science, mechanical engineering and industrial engineering and operations research people. The PATH research has consistently focused on the combination of cooperating vehicle and infrastructure elements to produce a wellintegrated transportation system, in preference to a more traditional practice of focusing on the separate vehicle or infrastructure elements. The large majority of the ITS community considered automation to be too futuristic to be worth their attention until very recently. However, it has consistently been a strong research theme at PATH since the beginning thirty years ago—indeed, it was one of the primary reasons for launching the program in the first place. During the late 1980s the PATH studies of automation were conceptual and theoretical, but in the early 1990s, they advanced into experiments with full-scale vehicles. PATH demonstrated high-performance automated

ConnectedAuto.org

Connected Auto

Fall 2016

steering control on a test track, using magnets embedded in the pavement as the primary position reference, in 1992, and demonstrated a four-car platoon under automatic longitudinal control at highway speed in 1994. These pioneering experiments under Caltrans sponsorship placed PATH at the international forefront in road vehicle automation and led to PATH and Caltrans being the only organizations represented on both of the teams that competed for the national automated highway system program in 1994. PATH was the most active of the core participants on the winning team, the National Automated Highway Systems Consortium (NAHSC).

During the late 1980s the PATH studies of automation were conceptual and theoretical, but in the early 1990s, they advanced into experiments with full-scale vehicles. In the NAHSC, PATH led the second stage of concept development and evaluation, the development of the modeling and simulation tools and the eight-car automated platoon demonstration that was shown to a large international audience at Demo ’97 in San Diego in August 1997. Nearly a thousand visitors received demonstration rides on these vehicles, which drove under fully automated control along the I-15 HOV lanes (when they were closed to public traffic), with maneuvers including lane changing and joining and splitting the platoon.

PATH recreated portions of this demonstration for Demo ’98 at Rijnwoude in the Netherlands and Demo 2000 in Tsukuba, Japan, in both cases as the only U.S. organization participating. All of these demonstrations showed the potential of SAE Level 4 automation on dedicated highway lanes to increase traffic throughput and reduce congestion while providing comfortable ride quality and cleaner and more efficient operations. After the termination of the NAHSC program, PATH developed automated bus and truck platoon demonstrations for Caltrans in 2003, showing how the same kinds of automation technology that were used on passenger cars in the NAHSC could be applied to heavy vehicles with the potential for earlier deployment. The truck experiments included carefully controlled measurements of the energy consumption and emissions reductions enabled by automation. The bus experiments led to PATH’s participation in the Federal Transit Administration’s Vehicle Assist and Automation (VAA) program, in which PATH equipped a transit bus with automatic steering control and precision docking capabilities. The lateral control system on this bus was refined to the level that it was placed in public revenue service for six months by Lane Transit District in Eugene, OR, where it was very well received by the agency and its drivers. This represented SAE Level 1 automation since the drivers maintained complete control of the speed of the buses. PATH’s truck platooning research continued under the FHWA Exploratory Advanced Research Program, where it was extended from two to three heavy trucks, including platoon join and split maneuvers and operations up and down grades. More recently, PATH has developed a cooperative adaptive cruise control (CACC) system for heavy trucks in collaboration with the Volvo Group, and is testing its energy saving potential in a combination of carefully controlled test track tests and on-the-road tests of truck driver preferences for different gap settings. This is also an SAE Level 1 automation system since the drivers maintain continuous responsibility for steering the trucks.

ConnectedAuto.org

PATH was the pioneer in the development of cooperative adaptive cruise control (CACC) for passenger cars since 2002. PATH was the pioneer in development of CACC for passenger cars since 2002, having already implemented three generations of CACC systems on three different passenger car platforms. This research included a human factors experiment to determine drivers’ preferences among the different vehicle following gap settings, which revealed a significant driver preference for the shortest available settings. These CACC experiments have produced unique data about CACC vehicle following dynamics and driver gap selection that PATH has incorporated into detailed micro-simulation models of CACC to predict the larger-scale traffic impacts of widespread use of CACC at different market penetration levels. Based on PATH’s extensive experience with road vehicle automation technology, it was recently chosen to serve as the technical advisor to the California Department of Motor Vehicles (DMV) for their work developing California’s regulations to govern the safe usage of automated vehicle systems (SAE Level 3 and above) on California’s public roads. In this regard, PATH has helped the DMV to understand the technology capabilities and limitations, to recognize a range of approaches for qualifying drivers, and to identify approaches for determining whether specific automated vehicles are ready to be permitted to operate on California’s public roads. PATH has recently announced the creation of the Berkeley DeepDrive (BDD) Industry Consortium. This research alliance is investigating state-ofthe-art technologies in computer vision and machine learning for automotive applications. The multi-disciplinary Center brings new faculty and researchers together from PATH, the Electrical Engineering and Computer Science Department, Center for Information Technology Research in the Interest of Society (CITRIS) robotics, and the Berkeley Vision and Learning Center (BVLC) to investigate stateof-the-art technologies merging computer vision and machine learning for automotive

applications. The Center has joined forces with 14 inaugural private industry partners.

In 2016, the Berkeley DeepDrive (BDD) Industry Consortium was founded to research state-of-the-art technologies in computer vision and machine learning for automotive applications. PATH is also conducting research on automated driving in urban environments on an international scale, with institutions such as the “Drive for All” foundation, headquartered at MINES ParisTech in France. The foundation has united researchers from UC Berkeley, MINES ParisTech, Shanghai JiaoTong University of China, and EPFL of Switzerland. The project aims at expanding knowledge of self-driving cars, developing intelligent on-board systems, and exploring deployment characteristics in different regions around the world. PATH is also collaborating with the Automotive Research and Testing Center (ARTC) and Industrial Technology Research Institute (ITRI) of Taiwan on vehicle automation. The collaborative research projects are focused on the studies of control transition between driver and machine in semiautomated systems, as well as the application of machine learning and robust control technologies for highly automated driving systems. Finally, PATH has been a primary fixture in the development of Connected Vehicle technology including the development of the nation’s first DSRC based test facility in Palo Alto, California and featured on page 42 of this publication. More recently, PATH researchers have developed new traffic signal control strategies using data from connected vehicles, including vehicle-to-vehicle and vehicle-to-infrastructure technologies. The new control strategies tested well and showed improved mobility and safety. Future articles will be dedicated to the details of PATH work in the area of connected vehicles and infrastructure.

For information regarding the PATH 30th Anniversary Gala Event to be held on February 17-18, 2017 on the UC Berkeley campus, please contact Thomas West, PATH Director at info@path.berkeley.edu.

ConnectedAuto.org

Connected Auto

Fall 2016

Seamless Location for Tomorrow’s Connected Car By Mike Mathews, Brian Salisbury, and Craig Peddie

ConnectedAuto.org

Cars are quickly becoming the most sophisticated and largest mobile devices in the Internet of Things (IoT). Facilitated by a variety of IoT technologies, drivers are becoming accustomed to enriched services and experiences, such as enhanced navigation, real-time traffic, parking information, and integration between dashboards, smartphones and wearable devices. Many of these services rely on one very crucial piece of information above all others—the car’s location.

ConnectedAuto.org

Connected Auto

Fall 2016

Car 54, where are you?

Connected Car Platform

Equipping cars with GPS, electronic compass, and wheel rotation sensors is nothing new. These signals have been used by navigation systems for many years, but have been primarily used outdoors. We’ve all experienced having our navigation system become confused when we drive through tunnels and urban canyons. With sensor fusion and hybrid positioning technology, the car’s location can be tracked without interruption.

Hybrid isn’t just for motors… Hybrid positioning is a method of combining multiple location technologies to yield a solution that is more accurate with higher-availability which is not independently achievable. Through statistical estimation, leveraging the positioning data from GPS, cellular, BLE, and Wi-Fi are used to simultaneously calculate a device’s position, speed, and heading. In addition, sensor fusion adds inertial sensor and geographical constraint data, which determine the relative changes in orientation and position. This is done all while adjusting for fixed geographic references such as roadway, parking spaces, walls, and other physical obstructions.

Connected Cars becoming part of the IoT Cloud

The position is continuously validated and corrected using any and all sources of information available. The benefits of hybrid and sensor fusion positioning are appealing. However, these techniques require significant assistance and contextual data in order to be effective. Partitioning hybrid positioning using a distributed architecture, where functional components are available both in the vehicle and the cloud, optimizes latency, data transfer, and performance. A hybrid positioning engine, like the one developed by Comtech Telecommunications Corp., can run locally embedded in the Connected Car Platform. This engine delivers real-time location information with virtually no latency and operating continuously without interruption—even when wireless connectivity is lost. However when connectivity is available, the local engine simply downloads any updates for GNSS and Network assistance data and location relevant contextual data.

ConnectedAuto.org

3D Position, Heading, and Speed Hybrid Positioning Sensor Fusion RF SLAM Cell ID / Wi-Fi AGNESS

Contextual Data Network Assistance Data GNSS Assistance Data

ILP API

Optional Direct Connect

with a single pass. This presents significantly higher performance than traditional crowdsourced Wi-Fi solutions, which average 30 to 50 meter accuracy in urban environments. Using the SLAM technique and hybrid positioning, connected vehicles can be positioned very accurately in GNSS obstructed environments and over time, the resulting system will continue to improve yielding better measurements of RF and the common paths traveled.

Finding that elusive parking spot

HTTP(S) JSON/Binary

Customer Application Gateway

Finding an open parking spot in multi-story and underground garages can be greatly simplified, when smart garages can communicate with connected cars. Indoor positioning solutions help guide the car to the closest suitable open spot.

HTTP(S) JSON/Binary

Location Studio™ IoT Location Platform (ILP) Services

Cloud

LBS Architecture for Connected Car

Comtech’s IoT Location Platform (ILP) provides cloud-based location services supporting the hybrid positioning engine including: Hybrid Positioning Calculation—Cloud based hybrid calculation, enables delegation to cloud when local calculation unavailable Network Assistance—Provides localized information about Wi-Fi hotspots and Simultaneous Localization and Mapping (SLAM) data

GNSS Assistance—Provides GNSS acquisition and tracking assistance data To further enhance accuracy and provide additional information about the environment, the hybrid positioning engine implements RF SLAM techniques, which enable precise mapping of the RF signatures of cellular, Wi-Fi, and BLE signals in the local environment. The resulting data is subsequently processed and used for positioning. This can greatly improve accuracy, achieving 3 to 5 meter accuracy

Newer parking garages are being built with occupancy sensors for individual parking spots. Currently, they are used for updating the signs at the garage entrance which advertise the number of empty spots. Inside the garage, a green light indicates if a spot is free, and red if it’s in use. This allows drivers to scan over the roofs of vehicles in the garage, looking for a green light and then drive in that direction. But why do you need to wait until you are in the garage and in sight of the empty parking spot? What we really want is for our navigation systems to guide us directly to an empty spot. Not possible, right? Wrong. The technology exists. Today.

IoT Location Platform

Next generation parking garage points or LoRa gateways may need to be installed in the garage, if ambient signals aren’t strong enough to provide coverage indoors. If the parking garage is private, the venue owner will want to keep the parking information private. In the case of a public garage, the information should be made available to all. With the Location Studio Context Content Server (CCS), the venue owner can decide who has access to their data. Location Studio based user experience for getting to that retail store or high rise office location now and it plays out like this: • The driver searches for their destination on a desktop, mobile or in the vehicle. All are sync’d seamlessly with Location Studio’s cloud navigation platform. • Navigation is started to the destination with the vehicle’s navigation system. The Location Studio routing engine alerts the driver to the fact that their final destination does not have parking immediately in front, and routes them to the nearest parking garage instead.

Location Studio™ from Comtech provides a suite of location technologies that includes position determination using a broad range of signals for any device. Cell-ID, GPS, Wi-Fi, BLE, and inertial sensors are all supported by this platform. Location Studio supports everything to not only enable a car, but to enable IoT as well.

• As the vehicle nears the parking garage, the mapping engine queries the CCS and loads the indoor map data for the parking garage, along with the real-time data for empty spots. The route is updated to guide the vehicle to the empty parking spot nearest to the driver’s final destination.

When a parking garage sensor system is installed, one additional parameter must be captured when each vehicle sensor is installed —which parking space it is monitoring. Next, a quick survey of the RF signals is required for Location Studio’s SLAM technology that can guide a driver to an empty spot. Wi-Fi access

• As the vehicle enters the garage and loses GPS coverage, position determination automatically switches over to Wi-Fi and inertial sensor data. This data, combined with Location Studio’s SLAM technology, accurately guides the driver to the empty spot. If equipped, the vehicle can even park itself.

ConnectedAuto.org

Connected Auto

• With the Location Studio cloud based navigation system, the remaining route to the driver’s final destination is automatically transferred to their mobile device. Continuing to utilize the Location Studio indoor positioning technology, the driver’s mobile device now guides them to their final indoor destination. This depends upon a mix of the signals available to the device, and the hybrid positioning engine to provide location. • Returning to the driver’s vehicle is a simple matter of tapping the ‘Take Me Back’ button on their mobile phone, and the needed audible and visual cues are provided to guide them back to their vehicle. • Once the driver arrives within visual distance of the vehicle, geo-fence alerts can be used to flash the vehicle lights or pulse the horn.

Data on the move Connected cars can share data such as slippery roads, traffic congestion, below freezing temperatures, heavy precipitation and low visibility to enable nearby vehicles and their drivers to be better prepared. Capturing the correct locations where these conditions exist is vital. The Location Studio platform enables data captured by vehicles to be correlated with road segments. Static information such as speed limits and turn restrictions are linked to road segments in the platform’s map database and used to calculate the optimal route for navigation. Additionally the dynamic information

Locating the empty parking spot

ConnectedAuto.org

captured by the vehicles currently traveling on those road segments becomes a source of real-time context that assists nearby drivers by alerting them to hazardous road conditions ahead. When windshield wipers are on, or traction control is triggered, these signals are available via the OBD-II interface in the vehicle. Multiple vehicles reporting the same signals in the same location provides the basis for flagging that road segment as wet and slippery. Tracking this data over time forms the basis for historical information that can be used when planning itineraries. For example, if a particular stretch of road is frequently slippery on certain days and times (perhaps due to irrigation run-off), this context is taken into account when comparing route options and estimating travel times.

Who’s behind the wheel? Autonomous vehicles are the subject of lots of attention these days. There will be many steps in the process leading to fully autonomous vehicles being ubiquitous. Each of these steps has a dependency on accurate and precise location being combined with other information to create the context that is needed to guide the autonomous car safely to its destination. Crowd sourcing of data by present day vehicles and/or the smart phones of their drivers, and correlating that data with high quality location readings is an important early step. Artificial Intelligence (AI) can be applied to video streams captured by vehicle cameras, or the smart phone camera when it is appropriately mounted. By processing these streams in real time in the vehicle, information regarding the vehicle’s surroundings can be derived. Lanes in the road can be recognized. The type of vehicle in the lanes ahead and adjacent can be recognized. Occupied and empty parking spots can be recognized. Road signs can be recognized and “read”. This information helps increase the quality and detail in both the “static” map database and the dynamic context, both of which are needed for autonomous vehicles to operate safely and share the road with human drivers. Interestingly, some of this information can be useful to human drivers as well, when properly presented in the instrument cluster of their connected car.

www.roboticdrives.com

RoboDrive motors from TQ: When precision steering counts The RoboDrive technology provides the highest power density at maximum torque range and overload capability in a compact design. ■

RoboDrive motors are available in a variety of sizes, from small (25mm) to large (115mm) and cover a torque range from .025Nm to 11Nm

■

Order as rotor and stator kits for design-ins, or as geared motors

■

Select from solid or hollow shaft models

Connected Auto

Fall 2016

In Palo Alto, the nation’s first dedicated short-range communication Test Bed is now open to test connected vehicle technology and applications. Map of California Connected Vehicle Testbed RSU locations

California

n 2005, Caltrans, the San Francisco Bay Area’s Metropolitan Transportation Commission (MTC), and the University of California’s PATH Program partnered to establish the nation’s first dedicated shortrange communications (DSRC) test bed in Palo Alto, California with the aim of assessing real-world implementations of connected vehicle technology and applications. This California Connected Vehicle Test Bed was recently made part of the United States Department of Transportation’s (USDOT) network of Affiliated Test Beds and now conforms to the latest technology standards and architecture of USDOT’s Connected Vehicle research program. Conveniently located near many of the automobile research labs, technology companies and industry start-ups in the Silicon Valley, the Test Bed is open and accessible to all industry partners.

Connected Vehicle

The California Connected Vehicle Test Bed is operated and managed by PATH and is supported by Caltrans, MTC and USDOT’s Intelligent Transportation Systems (ITS) Joint

Program Office (JPO). It supports cutting edge research in connected vehicle safety, mobility and environmental related applications and services. The Test Bed utilizes stateof-the-art 5.9 GHz DSRC devices that span 11 consecutive intersections along a 2-mile stretch of the highly travelled arterial of El Camino Real (SR-82) arterial in Palo Alto. All of the intersections broadcast signal phase and timing (SPaT) and MAP messages over DSRC and are connected to a 4G LTE backhaul. This provides an ideal living laboratory for industry partners interested in testing their connected vehicle applications. In addition, Caltrans has plans to expand the test bed to cover more intersections along El Camino Real in the near future. Stay tuned for more information on the California Connected Vehicle Test bed in an upcoming article of Connected Auto. For now, please contact Benjamin McKeever at California PATH (ben.mckeever@ berkeley.edu) for additional details on the California Connected Vehicle Test Bed.

Test Bed By Benjamin McKeever, PATH

ConnectedAuto.org

Connected Auto

Fall 2016

Impact of Automated Vehicle Technologies on Driver Skills By Center for Automotive Research (CAR) Michigan Department of Transportation (MDOT)

The evolution towards more automation and connectivity is one of the greatest forces driving the current changes in the automotive industry. In the history of automobiles, technological evolution has always been followed by a parallel evolution in driving skills and human-vehicle interaction. As automobiles become more automated and connected, it is likely that drivers will change their driving behaviors accordingly.

(Please see final page for individual contributors)

ConnectedAuto.org

Connected Auto

Fall 2016

New technologies used in automated vehicles (AV) will augment or perform more of the tasks normally required of human drivers. In the automotive industry, as well as the research and public sphere, it is widely believed that drivers will become dependent to some extent or even overreliant on the automation. However, drivers may still want to resume manual operation of the vehicle in some occasions. They may also be required to do so, when faced with an event beyond the programming of these technologies or an automation failure. It is therefore vital to study the implications of these situations, because if the system does not perform as the driver anticipates, this may create dangerous situations that would not exist in the absence of the AV feature. This two-part paper is based on a report that analyzes the impacts of technological dependence on driving skills and on the ability of an operator to resume manual operation of the vehicle, in the event of a system failure or limitation. The findings of this report indicate great challenges to come for the automotive industry and the regulatory authorities. Important progress needs to be made for example on human-machine interfaces of AVs, and on training and licensing drivers who will use these new vehicles. Part one will examine the evolution and development of the connected and automated vehicles, the levels of automation, the basic skills needed to monitor the normal functioning of automated driving systems, and respond to the failures and limitations of such systems. Part two will continue in the next issue of Connected Auto.

DEVELOPMENT OF CONNECTED AND AUTOMATED VEHICLES The auto industry has been envisioning self-driving cars since at least 1939, when General Motors presented its “Futurama” concept at the World’s Fair. This system used a combination of road-embedded magnets and radio communication to guide vehicles without driver control.

ConnectedAuto.org

Automated vehicle technology did not hit the consumer market until decades later and then only in limited applications. The earliest automated vehicle technologies to be widely deployed were conventional cruise control (CCC), electronic speed control (ESC), and anti-lock braking systems (ABS). In recent years, more advanced automated vehicle systems, such as automated park assist, adaptive cruise control (ACC), and automated emergency braking (AEB), have become available in an increasing number of vehicles. The most exciting prospect, however, is the fully automated, selfdriving vehicle. It is difficult to predict at this point how and when connected and automated vehicles (CAV) will be available for sale and will finally be adopted, even as increasing numbers of private and public sources publish their projections. Two distinct conceptions about the role of the automated driving systems coexist. Driving automation can be seen, on the one hand, as a way to substitute or, on the other hand, to augment the driver’s role. The first of these two opposite conceptions implies that machines are better at driving than humans and should therefore replace humans in the driving role. Consequently, there is no need to keep humans in the loop; they can focus on other activities while the vehicle is operating autonomously. The second view is that humans will work more efficiently and effectively if they are provided with powerful and fast tools. Therefore, the goal is to extend or augment human driving capabilities with intelligent machines. In this case, human operators must be informed and involved in the driving task at all times. This conception also implies that humans and automated vehicles must mutually monitor themselves and mutually communicate their intents as copilots. Nearly all technologies relevant to this report belong to one or more of three general categories: (1) Automated Vehicle Systems, (2) Connected Vehicle Systems, and (3) Intelligent Transportation Systems. Automated vehicle systems improve vehicle performance or driver convenience by automatically controlling vehicle actuation systems. Connected Vehicle Systems involve information flow between the vehicle and the world. Intelligent Transportation Systems (ITS) refers to intelligent infrastructure and system management. As shown in Figure 1, these systems, alone or in combination, enable valuable products and services.

Figure 1. Taxonomy of advanced transportation and vehicle technology For the purpose of this discussion, it is useful to understand the three basic functional components of

For the purpose of this discussion, it is useful to understand the three basic functional components of

automation, which are monitoring, agency, and action, as depicted in Figure 2. Monitoring can be automation, which are monitoring, agency, and action, as depicted in Figure 2. Monitoring can be For the purpose of this discussion, it is useful to understand while agency consists of decision-making, and action involves viewed as sensing and paying attention, while agency consists of decision-making, and action involves viewed as sensing and paying attention, while agency consists of decision-making, and action involves the three basic functional components of automation, which implementing decisions. Furthermore, automated systems implementing decisions. Furthermore, automated systems that are considered to be ‘intelligent’ also implementing decisions. Furthermore, automated systems that are considered to be ‘intelligent’ also are monitoring, agency, and action, as depicted in Figure 2. that are considered to be ‘intelligent’ also usually include usually include various feedback loops and possibly even machine learning. Monitoringusually include various feedback loops and possibly even machine learning. can be viewed as sensing and paying attention, various feedback loops and possibly even machine learning. IGURE 2. 2. G G ENERALIZED SYSTEM FFIGURE ENERALIZEDAUTOMATED AUTOMATED SYSTEM

Monitoring: Monitoring: Sensor Input Sensor Input Operator Input Operator Input Communication Systems Data Communication Systems Data

Agency: Agency: Stored Contextual Data Stored Contextual Data Signal Processing Signal Processing Data Fusion Data Fusion Decision Algorithms

Action: Action: Physical System Actuators

Physical System Actuators

Decision Algorithms

Feedback Loop

A relevant distinction for the current report can be made between driver assistance systems, or

Figure 2. Three basic functional components of automation

ConnectedAuto.org

Connected Auto

Fall 2016

A relevant distinction for the current report can be made between driver assistance systems, or advanced driver assistance systems (ADAS), on the one hand, and automated driving systems (ADS), on the other hand. ADAS do not assume all the aspects of the dynamic driving task, and thus require a human driver to be actively engaged at all times. ADS control all aspects of the dynamic driving task, implying that ADS-equipped vehicles are self-driving. Most automated vehicle systems available today are not coupled with connected vehicle or ITS technologies. However, the utility of automated vehicle systems could be improved if such systems were coupled to connected vehicle systems and/or ITS. For example, if two or more vehicles with ACC were able to communicate in real-time via dedicated short-range communication (DSRC), the vehicles could engage in cooperative adaptive cruise control (CACC), or automated platooning. The decrease in following distance allowed by CACC can relieve drivers of many aspects of the dynamic driving task and improve fuel efficiency by as much as 15 percent in certain scenarios. Researchers are also investigating the potential for CACC to increase highway capacity without expanding the physical infrastructure. The USDOT envisions a future transportation system with broad interdependencies between automated and connected vehicle systems with ITS infrastructure.

LEVELS OF VEHICLE AUTOMATION

Table 1. NHTSA levels of automation

DESIGNATION

Level 1 Function-specific Automation

NHTSA Preliminary Statement of Policy

•

SAE J3016

•

UK Department for Transport (DfT)

•

UK Parliamentary Office of Science and Technology (POST)

•

German Highway Research Institute (BASt)

One or more specific of the control functions are automated. If multiple functions are automated, they operate independently of each other. The driver has overall control and is solely responsible for safe operation. The driver can choose to cede limited authority over a primary control, the vehicle can automatically assume limited authority over a primary control, or the automated system can aid the driver in certain normal driving or crash-imminent situations. The automation system does not replace driver vigilance and does not assume driving responsibility from the driver. E.g.: conventional cruise control (CCC) and adaptive cruise control (ACC), electronic stability control (ESC), dynamic brake support, and lane keeping.

Level 2 Combined Function Automation

At least two primary control functions are automated and work in unison to relieve the driver of control of those functions. Driver and vehicle have shared authority in certain limited driving situations. The driver is responsible for monitoring the roadway and operating the vehicle. The driver is expected to be available for control at all times and on short notice. The system can relinquish control at any time with no advance warning. The distinction between Level 1 and Level 2 is that the driver may operate the vehicle without their hands on the wheel and foot off the pedal at the same time. E.g.: adaptive cruise control in combination with lane centering.

Various organizations have introduced taxonomies and classifications of automation ‘levels’ to differentiate between systems with various capabilities. Current taxonomies include the following: •

DESCRIPTION

Level 3 Limited Self-Driving Automation

Vehicles at this level of automation enable the driver to cede full control of all safetycritical functions under certain traffic conditions. In automated mode, the driver may rely heavily on the automated driving system (ADS) to monitor for changes in the driving environment that would require transition back to driver control. The ADS is expected to alert the driver that they must reengage in the driving task with sufficiently comfortable transition time (i.e., an appropriate amount of transition time to safely regain manual control). The driver is not expected to constantly monitor the roadway while driving. E.g.: a self-driving car that can determine when the ADS is no longer able to function, such as when approaching a construction zone.

Level 4 Full Self-Driving Automation

The ADS is designed to perform all safety-critical functions and monitor roadway conditions for an entire trip. The driver is not expected to be available for control at any time during the trip. Safe operation rests solely on the ADS. This includes both occupied and unoccupied vehicles.

In the United States, the levels introduced by NHTSA in the Preliminary Statement of Policy and by SAE International are of particular importance. While the NHTSA classification is preliminary and does not have the force of law, it is important because it reflects NHTSA’s potential approach to automated vehicles if/ when the agency adopts formal regulations. NHTSA’s levels of automation have become largely adopted by industry stakeholders as the de facto measurement of automated vehicle capability in the US.

ConnectedAuto.org

Connected Auto

Table 2. NHTSA VS. SAE Levels of Automation

NHTSA LEVEL Advanced Driver Assistance Systems (ADAS)

SAE LEVEL

Level 1 – Function-specific Automation

Level 1 – Driver Assistance

Level 2 – Combined Function Automation

Level 2 – Partial Automation

Level 3 – Limited Self-Driving Automation Automated Driving Systems (ADS)

Level 3 – Conditional Automation

Level 4 – High Automation Level 4 – Full Self-Driving Automation

Leading-Edge Automotive Research CAR forecasts industry trends, advises on public policy, and sponsors multi-stakeholder communication forums.

Level 5 – Full Automation

For consistency, this report uses only the NHTSA taxonomy in its analysis. The NHTSA levels of automation are more relevant for they discussion, as they focus more on driver interaction with automation. Parallel to these levels of automation control, there is a more important question concerning the level of authority assigned to the automation. This relates to where lies the ultimate authority (or the power of veto) over the vehicles actions, with the human operator (soft automation) or with the technology (hard automation). Hard automation employs technology to prevent human error, has ultimate authority on the vehicle and can override the human operator’s inputs. Soft automation on the other hand can be overridden by drivers if they want or need to.

RECENT RESEARCH Impact of New Mobility Services on the Auto Industry

Table 3. Hard and soft automation technologies

HARD AUTOMATION

SOFT AUTOMATION

Automatic transmission

Cruise control

Anti-lock braking system (ABS)

Adaptive cruise control (ACC)

Traction control

Automated steering (AS)

Electronic stability control (ESC)

Collision waning system (CWS)

Collision avoidance system (CAS)

Parking aids

Some of these technologies already have been implemented. Level 1 driver assistance systems can be as simple as ABS, ESC, and ACC. Several automakers have introduced automated driving systems (e.g., traffic-jam assist) that meet the requirements of NHTSA Level 2 automation. A few manufacturers have introduced systems that appear to straddle the line between Levels 2 and 3. Vehicles at this level of automation enable the driver to cede full operation of all safetycritical functions under certain traffic or environmental conditions, and in those conditions to rely heavily on the vehicle to monitor the driving environment. This would suggest Level 3 status, but these systems are not generally identified as Level 3 automation because the driver is expected to actively observe the system and remain vigilant for situations that require reengagement with the physical aspects of the dynamic driving task with minimal warning. It seems only a matter of degree of reliability that distinguishes Levels 2 from 3 in this case.

Global Harmonization of Connected Vehicle Communication Standards Effects of the 2017-2025 EPA/NHTSA GHG/Fuel Economy Mandates Costs of Lack of Harmonization of Safety Regulations Impact of Automated Vehicle Technologies on Driver Skills

www.cargroup.org 50

ConnectedAuto.org

Fall 2016

Photo/ Wikipedia Commons: Steve Juvetson, Author The Google prototype self-driving car is likely the most advanced of these projects. Google’s self-driving car has been prepared to navigate every street in Mountain View, CA, and may soon be made available in beta-version to select non-test-drivers for use, such as typical Google employees. However, none of the vehicles currently for sale would allow a person to safely and reasonably “let the car drive”. Automated low-speed shuttles have been deployed in a few locations, but only in very limited pilot projects.

CONNECTED AND AUTOMATED VEHICLE CONNECTED AND AUTOMATED VEHICLE TECHNOLOGIES TECHNOLOGIES

As stated in the introduction, a great variety of CAV technologies is As stated in the introduction, a great variety of CAV technologies is being developed and some of these being developed and some of these features are already in use. features are already in use. For the purpose of this report, it is important to understand the interactions For the purpose of this report, it is important to understand the interactions between the driver and the between the driver and the technologies and especially the situations in technologies and especially the situations in which these features activate. Table 3 presents a brief which these features activate. Table 3 presents a brief description of the latter description of the latter aspect, for some of the most important AV technologies. An understanding of aspect, for some of the most important AV technologies. An understanding the activation methods of these features is useful for grasping how a driver would react upon resuming of the activation methods of these features is useful for grasping how a manual control of the vehicle. driver would react upon resuming manual control of the vehicle. Several manufacturers are working on these technologies and for the moment, their solutions are quite diverse. For instance, there are no standard Forward Collision Warning (FCW) parameters such as gap Several manufacturers are working on these technologies and for the setting, type of warning display or modality (visual, haptic, or auditory). moment, their solutions are quite diverse. For instance, there are no standard Forward Collision Warning (FCW) parameters such as gap setting, type of warning display or modality (visual, haptic, or auditory). The distribution of responsibility between the driver and the automated driving system changes with the increase of the level of automation, as illustrated in Figure 4. The distribution of responsibility between the driver and the automated driving system changes with the increase of the level of automation, as illustrated in Figure 3. F IGURE 3. S PECTRUM OF AUTOMATION DEGREE BETWEEN DRIVER AND AUTOMATION CONTROL Higher level of automation

Examples:

Figure 3. spectrum of automation degree between driver and automation control

Automation control

Driver control

ACC

TJA

Platooning

ConnectedAuto.org

Connected Auto

Fall 2016

A complete overview of all potential limitations and failures of the aforementioned AV technologies is beyond the scope of this report. An extensive failure mode and effects analysis (FMEA) of the AV features described in Table 3 would be needed for a complete reliability study. A FMEA involves reviewing as many components, assemblies, and subsystems as possible to identify failure modes, and their causes and effects. This would then allow for an extensive analysis of drivers’ reactions and performances for each type of failure. For the purpose of this report, we have assumed that, hypothetically, any part of a automated driving system (as schematically described in Figure 4) could be susceptible to failure. Figure 4. Logical architecture of vehicle automation with examples of components Perception Camera Radar Lidar GPS

Decision and control

Decision making Sensor fusion

V2X

Actuation

Throttle Brakes

Control

SKILL SET REQUIRED FOR USING AUTOMATED VEHICLE TECHNOLOGIES A typical modern driver undoubtedly has a unique skill set compared to a driver from previous decades. While many of today’s drivers have never mastered a manual transmission, or have never been required to pump the brakes on a slippery road, they are beginning to utilize modern CAV features such as adaptive cruise control or blind spot warning lights, as well as touch-screen interfaces for infotainment functions and hands-free calling. The steady integration of these AV technologies into modern vehicles will continue to influence the skill set required to safely operate in traffic.

SKILLS NEEDED TO MONITOR THE NORMAL FUNCTIONING OF AUTOMATED DRIVING SYSTEMS

Steering Other

Other

For the purpose of this report, failures of automation levels 1 to 3 are divided into three general Figure 4. Logical architecture of vehicle categories:

automation with examples of components • Errors: the automation activates when it is not required; • Misses: automation fails to activate; For the purpose•of Partial this report, of does automation levels misses:failures automation not fully activate. 1 to 3 are divided into three general categories:

Advanced AV technologies that will potentially see deployment in the coming years and decades could imply the need for a drastically different set of driving skills than the one needed for today’s vehicles.

Driving AVs will require better supervision and selective intervention skills, rather than manual For example, several research projects mentioned in chapter 5 studied different types of ACC failures: unwanted acceleration, complete lack of deceleration, partial lack of deceleration, speed limit violation. control and maneuvering skills. The supervision role, especially for NHTSA automation levels 2 • Errors: the automation activates when it is not required; FCW systems also have, for the time being, potential failures. In picking up target objects, FCW features and 3, will require skills in terms of coordination are limited by the capacities of their radar system. For example, the radar may select static roadside • Misses: automation fails to activate; (sharing information), cooperation (being aware objects by mistake or, especially on a curved road, a target in an adjacent lane may be chosen as the of and supporting each other’s goals), and • Partial misses: automation does not fully activate. lead vehicle. By mistaking another target as the lead vehicle, a FCW is likely to provide false alarms. collaboration (working on a shared project). For example, several research projects mentioned in the report studied different types of ACC failures: unwanted acceleration, complete lack of deceleration, partial lack of deceleration, speed limit violation. FCW systems also have, for the time being, potential failures. In picking up target objects, FCW features are limited by the capacities of their radar system. For example, the radar may select static roadside objects by mistake or, especially on a curved road, a target in an adjacent lane may be chosen as the lead vehicle. By mistaking another target as the lead vehicle, a FCW is likely to provide false alarms.

ConnectedAuto.org

Additionally, drivers will need to adapt to the different levels of automation and to understand the distribution of tasks between automation and manual control for each level. In other words, familiarity with the electronic functions of AVs will be required for all drivers using automation levels 1 to 3. Furthermore, operators will need to know when and how to interact with automated driving systems.

In-depth analyses have shown that, from a human factors perspective, each of the automation levels 1, 2, and 3 require different skills from drivers, especially in terms of situational awareness. For example, highly automated driving (HAD), NHTSA Level 3 automation, on the one hand, and ACC driving (Level 1), on the other hand, require different types of interaction between the vehicle and the human operator. Namely, in a highly automated car, the driver has the possibility to perform complex secondary tasks, a feature that is seen as one of the biggest advantages of HAD, whereas in an ACC equipped car, the driver must attend to the roadway constantly. Therefore, the gradual deployment of automation will put more emphasis on progressive and continuous training, rather than the current, one-off, initial training. On a more general note, drivers will need to develop the capacity to maintain a constant level of awareness of the performance of the AV and the environment, while, at the same time, performing secondary tasks with a variety of difficulty and attention requirements. Overall, operators should also know when it is safe to engage in secondary tasks. Namely for NHTSA level 3 vehicles, drivers must develop the ability to resume manual operation of the vehicle in a timely manner when the system requires it, as they are not required to constantly monitor the roadway. Therefore, drivers will need to master the techniques to transition from automation to manual control, as well as between different levels of automation.

List of Abbreviations used in this article ABS

Anti-Lock Braking Systems

ACC

Adaptive Cruise Control

ADAS

Advanced Driver Assistance Systems

ADS

Automated Driving Systems

AEB

Automated Emergency Braking

Automated Steering

CCC

Conventional Cruise Control

CAR

Center for Automotive Research

CAS

Collision Avoidance System

CAV

Connected and Automated Vehicle

CWS

Collision waning system

ESC

Electronic Stability Control

FCW

Forward Collision Warning

IMA

Intersection Movement Assist

ITS

Intelligent Transportation Systems

MDOT

Michigan Department of Transportation

TJA

Traffic-Jam Assist

SAE ORAVS Society of Automotive Engineers (SAE) On-Road Automated Vehicle Safety (ORAVS)

Lead Author

Adela Spulber, Transportation Systems Analyst, CAR

Additional Contributors

Can’t wait for January? The full report, including references is available at: http://www.cargroup.org/?module=P ublications&event=View&pubID=139

Richard Wallace, M.S., Director, Transportation Systems Analysis, CAR Gary Golembiewski, Senior Research Associate, Leidos Matt Smith, ITS Program Manager, MDOT |Niles Annelin, Transportation Planner, Connected and Automated Vehicle Policy, MDOT

This document is a product of the Center for Automotive Research under a State Planning and Research Grant administered by the Michigan Department of Transportation.

ConnectedAuto.org

Connected Auto

Fall 2016

Using

FRAM

to Build Smart Airbag Automotive Applications By Harsha Venkatesh

Ferroelectric-RAM (FRAM) memory has a wide variety of applications, including industrial control systems, industrial automation, mission-critical space applications, and automotive systems. Safety systems for automobiles are expected to become more sophisticated over the next several years. A principal driver of this trend is expected regulation which will impact both the attach rate and the sophistication of airbags and stability control systems. This article looks at the key technological advantages of using FRAM non-volatile memory technology in these systems.

ConnectedAuto.org

Connected Auto

Fall 2016

FRAM is a memory technology well suited to this requirements. Like other alternatives, it provides reliable nonvolatile storage. The main advantages of FRAM are its very high write endurance and its write speed. With FRAM, systems are able to continuously store data at full bus speed without the need for additional memory or overhead to manage the memory’s endurance with techniques such as wear-leveling. This is because FRAM is instantly non-volatile, requiring no additional soak-time to store information. Its write endurance is also on the order of 1014. Compare this to most EEPROMs and FLASH which have an endurance of less than 106.

The second major advancement seen in the market is the need for collection of “actual information or data” just before an accident in an Event Data Recorder(EDR). This can be very valuable for future litigation / insurance claim related incidents. The EDR function is normally included in the airbag electronic control unit (ECU). This is a natural grouping because the EDR does not have the survivability requirements of an airplane “black box” and the airbag controller is the primary recipient of a variety of important sensor inputs. Vehicle makers are also quick to point out that there is no room for a stand-alone EDR.

Airbag Design Due to the demand for safety and the high cost of replacement, manufacturers have added a variety of additional sensors to monitor and record the person sitting in the seat. These include a passenger occupancy pressure sensor, which is used to enable the airbag subsystem, along with a variety of position sensors to boost the effectiveness of the airbag system. Position data is constantly being updated and must be stored up to the point and even after the point of system deployment. The requirement to constantly log position data and store it in non-volatile memory makes high performance, low-power and high endurance FRAM an ideal solution.

These two requirements lead to the need for robust non-volatile memory that has a high endurance and fast access. In the case of a “smart” airbag, the designer wants to deploy the airbags with a variable force upon impact. The requirement for the memory is to frequently log the actual seat positioning and FEATURE the weight and presence / actual SPI Speed position of the occupant. In the case of maintaining the history leading I2C Speed up to the impact – the memory should have enough capacity to Write Delay store the last 15–20 seconds of Write Endurance (Cycles) information in a rolling buffer of data-logger. Since a normal car Lifetime @10-ms Write Frequency is designed to work for 30+ years, this memory needs to have a fast Active Write Current write time, instant non-volatility, and very high endurance. Density Range (AEC-Q100 125C)

ConnectedAuto.org

FRAM

Drawbacks of Floating gate technologies like FLASH / EEPROMs

The main advantages of FRAM are its very high write endurance and its write speed.

High-Level Non-Volatile Memory Comparison

EEPROM

NOR Flash

25 MHz

10 MHz

75 MHz

3.4 MHz

1 MHz

N/A

0 ms

5 ms

10 ms

1013

106

105

3,171 years

2.78 hours

0.28 hours

5 mA

6 mA

15 mA

4Kb-2Mb

1Kb-1Mb

1Mb-256Mb

As the complexity of automotive design requirements increases, the restrictive nature of floating gate memory technology becomes more apparent. For example, the programming process for floating gate-based memory takes several milliseconds, which is an inordinately long time for safety-critical applications. In the kind of fast power outage that occurs in a crash, little information could be stored successfully in a floating gate device. The programming process is also destructive to the insulating layer and such devices consequently have limited write endurance of typically 100,000 to 1,000,000 writes. In an occupant sensor, for example, data is updated too often for this upper limit. Given a typical requirement to write data once per second, a floating gate device would wear out in less than twelve days of operation. Buffering the data in RAM and writing to a floating gate nonvolatile memory on power down introduces the speed problem that occurs in the EDR, and so is not a viable solution. In smart airbag systems, while it is not only necessary to store data in the event of a crash, it is also desirable to store pre-crash data prior to an event. Using a rolling log to store precrash data is ideal, but this approach proves problematic for floating gate memory devices because of their limited endurance. Since airbag modules have large capacitors which store sufficient energy to fire the airbag, there may be sufficient residual energy to write the data from a buffer after the squib has fired. The amount of data that can be written is limited by the energy available, that is to say the residual energy in the capacitor and the speed with which the memory can be written. A typical 2K byte floating gate memory device can write approximately 4 bytes or 5 ms. To write an entire floating gate memory device, therefore, can take more than a second.

To write an entire floating gate memory device, therefore, can take more than a second.

Typical Block Diagram of Airbag-Systems A Nonvolatile Data Buffer Because of its high write endurance, FRAM can be used as a data buffer. The MCU can continuously log events in FRAM directly during runtime. Since FRAM is an intrinsic nonvolatile memory, it saves data after the power loss. Therefore, last moment data is never compromised even if the main power supply fails catastrophically. Since data is directly written in FRAM, a last moment data transfer from SRAM to nonvolatile space such as EEPROM or Flash is not required. Use of FRAM requires ‘zero’ system power backup for retaining last moment crash data.

100,000,000

80,000,000 Millions

Two major changes are occurring in “Airbag” systems. First, all new airbags have a smart sensor that detects the presence or non-presence of a passenger in the car. Every airbag that ejects incorrectly results in a costly replacement with associated maintenance, labor and parts costs. A smart sensor that keeps monitoring the weight and presence of the occupant can add a measure of “variability” to the force with which an airbag operates. This could both prevent airbag related injuries as well as help protect the passenger during severe crashes.

60,000,000

40,000,000

20,000,000

F-RAM

EEPROM

NOR FLASH

Cypress’s FRAM Write Endurance Vs EEPROM and Flash Write Endurance

ConnectedAuto.org

Connected Auto

No Write Delays Some events need to be recorded as often as 100 to 1000 times a second to capture every detail. However, this is a key challenge for existing EEPROM- and Flash-based recorders. EEPROM stores data on a page-by-page basis, and it requires a few milliseconds store time delay between two page writes into EEPROM. This limits data logging capability. ’No Delay’ writes in FRAM allow a system designer to capture and write real time data at the system bus speed.

Fast Writes and Low Power Consumption High-speed serial SPI and I2C interfaces and/ or high-speed parallel synchronous access in FRAM enable a controller to write data in FRAM in a less time due to the best in class speed specifications for nonvolatile writes. Low power FRAM also requires only a fraction of the total power required by other nonvolatile memory technologies.

’No Delay’ writes in FRAM allow a system designer to capture and write real time data at the system bus speed.

12000000

High Reliability Data reliability of the EDR is important to attain the goals of accuracy, survivability, data retrieval, and most importantly, durability. Since this memory space is used for logging critical sensor data, high reliability and data integrity are a must for automotive applications.

Maturity Technology maturity is a larger concern for the automotive market than for other types of applications. EEPROM and Flash technologies are well-understood, with established quality control infrastructures at major suppliers. The introduction of new technology naturally causes hesitation, as the technical community must become comfortable with its reliability and availability. With over 500 million units shipped in automotive environment (also in under-hood applications at extreme temperature grades of 125°C), FRAM has matured to the point where automotive customers can feel comfortable and confident.

With over 400 semiconductors per vehicle expected in 2017, it’s no wonder that computer companies have started making cars

FRAM can lower system cost, increase system efficiency, and reduce complexity while being significantly lower power than Flash, EEPROMs, battery-backed SRAMs, and other comparable technologies.

10000000 8000000 6000000 4000000 2000000 0

F-RAM 314

EEPROM 63,996

NOR FLASH 10,611,178

Cypress’s FRAM Write Energy vs EEPROM and Flash Write Energy

Data reliability of the EDR is important to attain the goals of accuracy, survivability, data retrieval, and most importantly, durability. Smart Airbag

ConnectedAuto.org

Connected Auto

Fall 2016

Designing Secure

INTERFACE SOLUTIONS

for Connected Vehicles

The best way to summarize the intent of this article is to borrow one of the rules of rhetoric first proposed by Aristotle in 350 B.C. In short, he suggested that you start by “telling them what you are going to tell them.”

And so we will. This article will present NXP’s secure interfaces and power solutions for automotive applications, which is part of NXP’s Secure Interfaces and Infrastructure (SI&I) business. This group serves the diverse and fragmented interface and system management sector with a broad portfolio including Bridge devices, I2C, SPI, LED-lighting controllers, realtime clocks, analog switches, level shifters and display port multiplexers, to name just a few. In total NXP’s SI&I has a portfolio of more than 700 products and more than 10,000 customers worldwide. The company also has delivered more than 200 million units to the

By Murray Slovic for Convergence Promotions LLC

ConnectedAuto.org

automotive industry to date.

ConnectedAuto.org

Connected Auto

Fall 2016

Automotive Interface Solutions Automotive Interface Solutions LED CONTROLLERS

REAL TIME CLOCKS

LCD DRIVERS

TEMPERATURE SENSORS

GPIO EXPANDERS

ANALOG SWITCHES

LOAD SWITCHES

CAPACITIVE SENSORS

DP-LVDS BRIDGE

LEVEL SHIFTERS

I2C-BUS ENABLERS

COMPARATORS

Figure 1. Secure Interfaces and Power Solutions for Automotive Applications

While the automotive-qualified secure interface www.nxp.com/interface4auto products depicted above in Figure 1 provide essential functions in a variety of automotive systems, for clarity it should be understood that despite the name (secure interfaces) they are not directly part of NXP’s best-in-class security products protecting against data theft and/or unauthorized access (which are covered by NXP’s Secure Connected Devices group). Figure 1 shows many of the categories of devices that we will be discussing. Our descriptions will concentrate on solutions and specific example parts, as well as key features and important specifications.

ConnectedAuto.org

USB TYPE-C

Real-time Clocks Specifically designed for automotive applications—such as instrument clusters, battery management units and car radios— NXP’s automotive real-time clocks are designed for high reliability and low-power consumption. Some use a 32.768kHz oscillator and one oscillator capacitor, and the only additional external components required are a 32.768kHz quartz crystal and a capacitor connected to the OSCI pin. Others, like the PCF2127 and PCF2129 do not require any external crystal. The PCF2123, PCF8523, PCF85063A, PCF85063B, PCF85263A, PCF85363A, PCF2127, PCF2129, and several other RTC devices do not require

any external capacitor on the OSCI and/or OSCO pin. Devices are available with I2C or SPI bus interfaces and provide year, month, day, weekday, hours, minutes and second information as well as feature programmable alarm and timer functions with interrupt capability. They have an extended operating temperature range of up to 125 °C and AEC-Q100 automotive compliant qualification. The new PCA85063A (Figure 2) is a RealTime Clock (RTC) and calendar with an offset register that allows fine-tuning of the clock. All addresses and data are transferred serially via the two-line bidirectional I2C bus. Maximum data rate is 400 kbit/s. The register address is incremented automatically after each written or read data byte. It features low current (typical 0.25 μA at VDD = 3.0 V and Tamb of 25 °C).

DisplayPort Bridge DisplayPort to LVDS bridge devices enable connectivity between a DisplayPort (eDP) source and LVDS display panel. It processes the incoming DP stream, performs DP to LVDS protocol conversion and transmits the processed stream in the LVDS format. NXP’s PTN3460I (Figure 3) has two high-speed ports: a receive port facing the DP Source (for example, a CPU/GPU/chip set), and a transmit port facing the LVDS receiver (for example, the LVDS display panel controller). The PTN3460I can receive a DP stream at a link rate of 1.62 Gbit/s or 2.7 Gbit/s and it can support 1-lane or 2-lane DP operation. It also supports single bus or dual bus LVDS signaling with color depths of 18 bits per pixel or 24 bits per pixel and pixel clock frequency up to 112 MHz. PTN3460I offers high flexibility to optimally fit under different platform environments. It supports three configuration options: multi-level configuration pins, DP AUX interface, and I2C bus interface.

OSCO OSCI

32 kHz OSCILLATOR

DIVIDER

POWER-ON RESET

CLOCK CALIBRATION OFFSET

VDO

SYSTEM CONTROL

VSS SDA SCL

I2C-BUS INTERFACE

CLOCK OUT

CLKOUT

INTERRUPT CONTROL

INT

REAL-TIME CLOCK ALARM AND TIMER CONTROL

PCA85063A aaa-013712

Figure 2. Block diagram of PCA85063A

Notebook or AIO Platform

CPU/GPU/ Chipset

DP - LVDS LVDS Bridge

LVDS panel Cable

Figure 3. PTN3460 context diagram The device can be powered by either a 3.3 V supply only or dual supplies (3.3 V / 1.8 V) and is available in the HVQFN56 7 mm x 7 mm package with 0.4 mm pitch.

Automotive LED Controllers NXP’s automotive LED controllers offer high reliability combined with LED dimming, blinking and color mixing capabilities. The controllers feature sixteen independent PWM channels for individual brightness control of each channel. These are complemented by one master PWM controller for global dimming of all sixteen channels. For asynchronous control of the LEDs, an

ConnectedAuto.org

Connected Auto

active low output enable pin allows engineers to program all the PWM outputs to logic ‘1’, logic ‘0’ or ‘high-impedance’. PWM resolution ranges from 8-bit (256 steps) with fixed 97 kHz PWM frequency to 12-bit (4096 steps) with programmable PWM frequency between 40 Hz and 1000 Hz. A two-wire I2C bus allows the automotive LED controllers to interconnect to the microcontroller at speeds up to 1 MHz (Fastmode plus or FM+). They are at 5 V capable to sink 25 mA respectively to source 10 mA per PWM channel. Qualified compliant to AEC-Q100 automotive qualification standard, the controllers are available in TSSOP28 package and offer the highest reliability.

SCL

Planet e: Where the future begins. Embedded solutions of tomorrow. Today. Embedded exhibition sector: November 8–11, 2016

The LED output driver is programmed to be either open-drain with a 25 mA current sink capability at 5 V or totem-pole with a 25 mA sink, 10 mA source capability at 5 V. The PCA9635 operates with a supply voltage range of 2.3 V to 5.5 V and the outputs are 5.5 V tolerant. LEDs can be directly connected to the LED output (up to 25 mA, 5.5 V) or controlled with external drivers and a minimum amount of discrete components for larger current or higher voltage LEDs.

The PCA9635 (Figure 4) is an I2C-bus controlled 16-bit LED driver optimized for Red/Green/ Blue/Amber (RGBA) color mixing applications. Each LED output has its own 8-bit resolution

Connecting Global Competence

(256 steps) fixed frequency individual PWM controller that operates at 97 kHz with a duty cycle that is adjustable from 0% to 99.6% to allow the LED to be set to a specific brightness value. An additional 8-bit resolution (256 steps) group PWM controller has both a fixed frequency of 190 Hz and an adjustable frequency between 24 Hz to once every 10.73 seconds with a duty cycle that is adjustable from 0% to 99.6% that is used to either dim or blink all LEDs with the same value.

electronica Embedded Forum: November 8–11, 2016 Embedded Platforms Conference: November 9–10, 2016 All about electronica Embedded: electronica.de/en/embedded

INPUT FILTER

SDA

I2C-BUS CONTROL POWER-ON RESET

VDD

VSS

LED STATE SELECT REGISTER PWM REGISTER X BRIGHTNESS CONTROL

97 kHz 25 MHz OSCILLATOR

24.3 kHz

LEDn

GRPFREQ REGISTER 190 Hz

MUX/ CONTROL GRPPWM REGISTER “0” – permanently OFF “1” – permanently ON

Tickets & Registration: electronica.de/en/tickets

Figure 4. Block diagram of PCA9635

002aacf36

World’s Leading Trade Fair for Electronic Components, Systems and Applications

ConnectedAuto.org

Messe München I November 8–11, 2016 I electronica.de

Fall 2016

Experience Real Design Freedom

LCD Display Drivers NXP’s Automotive LCD Segment Drivers are designed for high reliability and optical performance. Devices are available for a wide range of resolutions from 4x18 up to 4x160 segments. There are cased and also have Chip On-Glass versions available. NXP’s Automotive LCD segment drivers are specifically designed for automotive applications such as instrument cluster, car radios and climate controls. Due their low power consumption these products are also suitable for key fob application. They offer all the necessary driving signals to interface to almost any low multiplex Liquid Crystal Display (LCD) of Twisted Nematic (TN), Super Twisted Nematic (STN) and Black Nematic (BN) type. NXP’s LCD segment drivers cover a wide range of resolutions from 4x32 up to 9x102 segments.

Only TQ allows you to choose between ARM®, Intel®, NXP and TI • Off-the-shelf modules from Intel, NXP and TI • Custom designs and manufacturing • Rigorous testing • Built for rugged environments: -40°C... +85°C • Long-term availability • Smallest form factors in the industry • All processor functions available

For more information call 508 209 0294 www.embeddedmodules.net

The PCA2117 is a low-power LCD controller and driver. It is specifically designed to drive LCD character displays of 2 lines by 20 characters or 1 line by 40 characters with 5 x 8 dot per character format. In addition, 200 icons can be driven by the PCA2117. The chip contains a character generator and displays alphanumeric characters. The PCA2117 features an internal charge pump with internal capacitors for on-chip generation of the LCD driving voltage. To ensure an optimal and stable contrast over the full temperature range, the PCA2117 offers a programmable temperature compensation of the LCD supply voltage. The PCA2117 can be easily connected to a microcontroller by either the two-line I2C-bus or a four-line bidirectional SPI-bus.

GPIO Expanders Used in body control units, instrument clusters, engine controls and car infotainment systems NXP’s GPIO Expanders provide reliable IO expansion for most microprocessor families, allowing designers to save GPIOs on microprocessors for other important functions. With the requirement for more functionality and features such as LED control, push button input control and system monitoring,

the relatively small numbers of GPIOs on microprocessors are becoming a limiting factor. Adding IO expansion with NXP’s GPIO Expanders overcomes these limitations. GPIO Expanders are available with either I2C bus or SPI bus interface and offer from 8 to 16 IOs. Some of the products feature an additional INT (INTerrupt) output and/or a REST or OE (Output Enable) input. The output is used to signal the microcontroller when any of the inputs change state (1 to 0, or 0 to 1). The RESET input can be used to initialize the device to its default state without cycling power to it. The AEC-Q100-compliant PCA9538PW/ Q900 and PCA9539PW GPIO for I2C bus/ SMBus applications are available as 8- and 16-bit GPIO Expanders with interrupt output and hardware reset input. These interface devices are rated for 4.5 V to 5.5 V operation. They offer design engineers using the I2C bus the flexibility of additional I/O port access without the need to replace microcontrollers. Additionally, NXP’s I2C bus/SMBus-based GPIOs enable seamless migration to newer microcontrollers while allowing design engineers to keep the same peripherals.

Voltage Level Translators NXP offers a range of bidirectional noninverting level shifter and translator circuits that meets the requirements of the automotive electronics council’s AEC-Q100 and recommended for automotive applications. Target applications include display clusters, infotainment, vision control and I2C. Bidirectional voltage level translator circuits are used to interface between applications with different supply voltage and input-output voltage levels. NXP’s bidirectional level translator options include variants of bit width ranging from single bit to 32-bit. Translators are available in a range of families providing bidirectional translation between 0.8 V to 5.5 V.

ConnectedAuto.org

Connected Auto

Fall 2016

Translators are available in standard logic and innovative PicoGate packages. Smaller leadless MicroPak, WLCSP and DQFN packages are available for PCB space saving. NXPautomotive translators are fully specified from -40°C to 125°C, and are qualified per AEC-Q100 grade 1.

Load Switches NXP’s Integrated Load Switches offer the best solution for automotive applications including power management systems, ABS, airbag, and other safety and comfort systems. Combining a PNP low VCEsat transistor with a RET they have a low threshold voltage (<1 V compared to MOSFET).

Comparators NXP offers a range of small footprint low power voltage comparator solutions for power sensitive and space constrained applications. The comparators are available in single and dual configurations with standard push-pull as well as open-drain outputs. Comparators are offered in PicoGate and leadless MicroPak packages for PCB spaces saving. NXP’s comparators are fully specified from -40°C to 85°C.

The NCX2200 (Figure 5) is a single low voltage low power comparator with a very low supply current of 6 μA. It operates at a low voltage of 1.3 V and is fully operational up to 5.5 V which makes this device convenient for use in both 3.0 V and 5.0 V systems. ESD protection includes HBM JESD22-A114F Class 3A (exceeds 2000 V) and CDM JESD22-C101E (exceeds 1000 V).

Temperature Sensors Combining accuracy, reliability and stability, NXP’s silicon temperature sensors are the ideal choice for automotive applications from climate control to engine monitoring. An extensive selection of operating ranges, packages, resistances and tolerances ensures designers can find the perfect solution to their temperature monitoring needs.

Analog Switches

NCX2200 OUT

VCC

VEE

n.c.

IN+

IN-

001aan853

Transparent top view Figure 5. Pin configuration for SOT886

NXP’s comprehensive range of analog switches include SPST to SP16T options that are suitable for a variety of automotive analog and digital switching applications including sample and hold circuits, sensor data multiplexing, video & audio switching and GPIO expansion. Analog switches are offered in TSSOP, PicoGate and leadless MicroPak and DQFN packages for PCB space saving. NXP’s automotive switches are fully specified from -40°C to 125°C, and are qualified per AEC-Q100 grade 1.

Capacitive Sensors By enabling human interfaces that can be controlled by proximity or touch, capacitive sensors transform the way we interact with electronic systems. Capacitive sensors work by using the human body as one of the capacitive plates of an electrical circuit. A leader in low power capacitance touch sensors, NXP offers a range of auto-calibrating sensors that compensate for changes in the environment, such as varying humidity or contamination on the electrode. Dirt, humidity, freezing temperatures or damage to the electrode do not affect the device’s function. Compared to alternative switches and sensors, capacitive options have a number of

ConnectedAuto.org

advantages. Resistance-based sensing relies on contact, but capacitive sensors can switch without contact due to proximity sensing. What is more, without the need for foil on the switching area, materials costs are reduced. The PCA8886 is a low power dual channel capacitive proximity switch that uses a patented (EDISEN) digital method to detect a change in capacitance on remote sensing plates. Changes in the static capacitance (as opposed to dynamic capacitance changes) are automatically compensated using continuous auto-calibration. Remote sensing plates (for example, conductive foil) can be connected directly to the IC or remotely using a coaxial cable. It is suitable for proximity sensing in door locks and grips as well as a dashboard switch to toggle menus and reset trip counters.

Summary As we have seen in this overview, NXP secure interface products can complement your design by providing essential interface, I2C, analog or RF solutions. For more information please visit: www.nxp.com/interface4auto

ConnectedAuto.org

Connected Auto

Fall 2016

Integrated Corridor Management Creates Safer and More Efficient Transportation Systems through Connectivity “Connected Corridors links travelers, transportation managers, and even vehicles, to the intelligence they need to make good decisions.” —Joseph Butler, Connected Corridors Program Manager

ConnectedAuto.org

Connected Corridors is a collaborative program to research, develop, and test an Integrated Corridor Management (ICM) approach to managing transportation corridors throughout California. Rather than focusing on improving only specific elements such as freeways or transit, ICM views the corridor as a total system to be managed as a connected and cohesive whole; it seeks to address the corridor’s overall transportation needs rather than the needs of particular elements or agencies alone.

ConnectedAuto.org

Connected Auto

Fall 2016

onnected Corridors represents a significant departure from traditional transportation management practice, and in pursuing an ICM approach, the program aims to fundamentally change the way the State of California manages its transportation corridors. The program is led by the California Department of Transportation (Caltrans) with assistance from UC Berkeley Partners for Advanced Transportation Technology (PATH).

ConnectedAuto.org

Connected Corridors is starting with a pilot deployment on Interstate 210 in the San Gabriel Valley near Los Angeles and will then expand to multiple corridors throughout California in the coming years. Program leaders are specifically developing the Pilot with replicability in mind, giving extra thought as to how to best implement ICM, and documenting every step along the way. “We are following a systems engineering approach, but we’re refining the process to make

it work throughout the state and for the specific stakeholders in each corridor,” says Joe Butler, Program Manager for Connected Corridors. For decades, individual agencies only dealt with traffic within their jurisdiction, but trips do not often begin and end in only one city or without the use of a state-owned freeway, especially in car-centric LA. By sharing data and managing traffic at the corridor level,

all trips will be optimized regardless of their specific jurisdiction. Bus and rail information will also be incorporated to provide better transit mobility and encourage mode shift. This is a holistic approach with the core belief that better data results in better systems and better decisions. “Everything is intelligent, but intelligent decisions can only be made with good data,” adds Butler. “Connected Corridors

ConnectedAuto.org

Connected Auto

Fall 2016

is focused on producing and distributing accurate and timely data, and using that data so decision makers and travelers can make better decisions and thereby produce a more efficient transportation system.” As connected vehicle technology progresses, so will the capabilities of this approach. Infrastructure and vehicles will be able to inform each other with better data being disseminated more rapidly through the connected vehicles and into the hands of transportation managers and the decision support system (DSS). Connected Corridors will create an avenue for new data to quickly be incorporated into management decisions including metering and signal timing plans, incident and event re-routing, and transit signal priority. Additionally, travelers will receive better data through CC information systems, changeable message signs, connected infrastructure, as well as their connected vehicle.

The I-210 Pilot Location The Pilot corridor located on Interstate 210 will extend from SR-134 to Foothill Blvd/ US-66, approximately 20 miles in length. Due to the magnitude of the pilot, it will be divided into two phases with the first phase extending from SR-134 to I-605. The I-210 corridor includes many attributes that are ideal for a pilot demonstration: • Well instrumented with roadway sensors • Directional traffic flow corresponding to morning and evening commute hours

ICM Strategies for the Pilot

The Importance of Partnership

All ICM strategies were reviewed and then selected by stakeholders based on the needs of the corridor and available funding. Strategies currently included are the integration of freeway ramp meters and arterial signal systems; arterial signal coordination; traffic re-routing due to incidents or events; transit signal priority; traveler communication (via changeable message signs, 511, radio, social networks, mobile app) of traffic conditions, transit services, alternate route/trip/mode options; and system coordination between Caltrans (freeway operator) and local jurisdictions (arterial and transit operators). Additional strategies being considered for future implementation include parking management and demand management with active participation from commuters and service providers.

Reaching the vision of connected vehicles, infrastructure, travelers, enhanced decision support, and social networking requires extensive collaboration with the people that manage the varying technologies and transportation assets in the corridor. Stakeholder participation has been prioritized since the very early stages of the Pilot and all major documents are reviewed and approved by the stakeholder group. Private sector partnerships are also proving valuable with their unique skill sets and innovative technologies. Current partners include AT&T, HERE, INRIX, and discussions are underway with additional companies. More private sector partners are needed to accomplish the Pilot’s ultimate vision.

Strategies that focus on connecting travelers, transportation managers, and vehicles with better data will be given priority. However, greater connectivity can also mean greater risks to the system. Advanced security measures and backup systems need to be in place to protect the system from hackers or outages. Data needs to be regularly verified to confirm its accuracy. Communication needs to be ongoing to ensure the system is working properly and is meeting the needs and expectations of the various stakeholders. Measures to reduce system vulnerabilities, the possibility of bad data, and other risks are incorporated into the Pilot’s planning and design documents and regularly discussed with stakeholders.

Connected Corridors is shifting the mindset of the industry towards integration and technology to optimize a transportation system and connecting vehicles, infrastructure and people with the information they need to make intelligent decisions. “Connectivity will become the standard for transportation systems and

The Connected Corridors Vision The Connected Corridors Vision Connected Corridors-VIP Vehicles, Infrastruture and People

Connected & Automated Vehicles

Connected Infrastructure

Connected Travelers

Enhanced Decision Support

Corrirod Centric Social Networking

information,” says Butler. “We are starting with data and the coordination of systems to achieve that connectivity, but for a truly connected world, this is only the beginning.”

The Connected Corridors Vision

• Substantial network of parallel arterials and a congested freeway • Existing traffic management infrastructure • Extensive transit service

ConnectedAuto.org

Connected Auto

Automotive Interface Solutions

I-210 Pilot Project Milestones The Connected Corridors team is following the Systems Engineering (SE) process, also referred to as the Vee Diagram. The team is currently in the system definition and design phase. Additional milestones include: • Stakeholder outreach and engagement - Ongoing • Concept of Operations complete - June 2015 • Project Charter executed - June 2015

LED CONTROLLERS

• System Requirements finalized – October 2016

“ICM seeks to optimize the use of existing infrastructure assets, making transportation investments go farther.”

REAL TIME CLOCKS

• 450 Intersection micro simulation complete – 2017 • Cloud-based data hub – Late 2017

LCD DRIVERS

• Project Demonstration Phase I – 2018 CREDITS: For more information: http://connected-corridors.berkeley.edu/

TEMPERATURE SENSORS

GPIO EXPANDERS

ANALOG SWITCHES

US Department of Transportation

Phase 0

Interfacing with Planning and the Regional Architecture

Regional Architecture

Concept Exploration Project Planning and and Concept of Operations Benefits Analysis Development

Needs Assessment

Concept Selection

Trade Studies Technical Reviews

Decision Gates

itio

Process Improvement

fin

Configuration Management

Program Metrics

System Definition and Design

System Development and Implementation

Validation, Operations and Maintenance, Changes & Upgrades

System Retirement / Replacement

Operations and Maintenance System Validation / Strategy Plan System Verification Plan (System Acceptance)

High-Level Design

nd na

Risk Management

o siti

Project Management Practices

Phase 5

System Requirements

Elicitation

Phase 4

Concept of Operations

com

Stakeholder Involvement

Phase 3

Systems Engineering Management Planning

Project Planning

Cross-Cutting Activities

Phase 2

Phase 1

Subsystem Requirements Detailed Design

Sub-system Verification Plan (Subsystem Acceptance)

System Validation Initial Deployment

Retirement / Replacement

USB TYPE-C

Decision Gate

Life Cycle Time Line

www.nxp.com/interface4auto ConnectedAuto.org

I2C-BUS ENABLERS

COMPARATORS

Traceability

DP-LVDS BRIDGE

LEVEL SHIFTERS

System Verification System Integration

Subsystem Verification Unit/Device Subsystem Test Integration Plan Unit Testing

Software Coding Hardware Fabrication

Changes and Upgardes

CAPACITIVE SENSORS

Int egr atio na nd Rec om po siti on

Phase -1

LOAD SWITCHES

Connected Vehicles. Connected Infrastructure. Connected Auto.

Connected Signals is a high-tech startup based in Eugene, Oregon that collects realtime traffic signal data and uses it to build sophisticated models for predicting traffic signal changes, based on history, vehicle and pedestrian calls, and other information. This information enables applications that: 1. increase safety at intersections, 2. improve fuel economy and reduce emissions, and 3. reduce driver stress.

Major automakers, including BMW and Toyota, estimate that 7â&#x20AC;&#x201C;10% improvements in fuel efficiency and emissions can be achieved using this data.

Live and predicted signal state in Portland, Oregon

At the juncture of big data, connected vehicles/mobile, and the Internet of Things, we use existing infrastructure to exchange data with traffic management systems and vehicles, without the need for the expensive municipal and in-vehicle equipment required by other connected vehicle approaches. With signal data from over 10,000 intersections in over 100 US municipalities, Australia, New Zealand, and Europe already online and many more in various stages from acquisition to deployment, Connected Signals is the leading source of real-time, predictive, traffic-signal data, providing added value to manufacturers, drivers, and local traffic agencies.

Connected Auto is your channel to reach the engineers, companies and auto manufacturers who are designing connected cars. Our newsletters,magazines,website and conferences are valuable marketing channels for companies selling products to this market. Make sure Connected Auto is on your radar for 2017.

Our EnLightenÂŽ smartphone app already provides tens of thousands of drivers with predictions about how long they will be stopped at lights, and helps them safely arrive at green lights. Through our collaborations with automotive OEMs and Tier 1 suppliers, as well as with other key players in the connected vehicle space, we are actively developing exciting new applications for connected and autonomous vehicles, including our recent deployment in BMW infotainment units in selected cities. Safety is our first priority, and we work carefully with our municipal, OEM, and other partners to ensure that drivers are presented with information in ways that are not distracting and help make drivers safer.

With the help of our proprietary V2If appliance, which we provide at no cost to traffic agencies, we are rapidly expanding our signal coverage across the United States and internationally. The V2If device, only one of which is required per traffic management system, provides a secure, one-way data feed without adding any load, or requiring software to be installed, on agency systems.

541-654-5829 355 Goodpasture Island Road, Suite 200

Eugene, OR 97401

541-654-5839 FAX

ConnectedAuto.org

Automotive Interface Solutions

LED CONTROLLERS

REAL TIME CLOCKS

LCD DRIVERS

TEMPERATURE SENSORS

GPIO EXPANDERS

ANALOG SWITCHES

LOAD SWITCHES

CAPACITIVE SENSORS

DP-LVDS BRIDGE

LEVEL SHIFTERS

I2C-BUS ENABLERS

COMPARATORS

USB TYPE-C