Engineering Magazine Spring 2024

CARNEGIE MELLON ENGINEERING

While the potential of new technologies is vast, it comes with responsibilities.

We need to build guardrails and designate guardians before releasing AI models to the public.”

ASSISTANT PROFESSOR GABE GOMES

Gabe Gomes (r.) with Robert MacKnight

AI COSCIENTIST AUTOMATES SCIENTIFIC DISCOVERY

Coscientist is a copilot for science, making experimentation faster and more accurate

Anon-organic intelligent system has for the first time designed, planned, and executed a chemistry experiment, Carnegie Mellon University researchers reported in the Dec. 21, 2023, issue of the journal Nature.

“We anticipate that intelligent agent systems for autonomous scientific experimentation will bring tremendous discoveries, unforeseen therapies and new materials. While we cannot predict what those discoveries will be, we hope to see a new way of conducting research given by the synergetic partnership between humans and machines,” the CMU research team wrote in their paper.

The system, called Coscientist, was designed by Assistant Professor of Chemical Engineering and Chemistry Gabe Gomes and chemical engineering doctoral students Daniil Boiko and Robert MacKnight. It uses large language models

(LLMs), including OpenAI’s GPT-4 and Anthropic’s Claude, to execute the full range of the experimental process with a simple, plain language prompt.

For example, a scientist could ask Coscientist to find a compound with given properties. The system scours the Internet, documentation data and other sources, synthesizes the information and selects a course of experimentation that uses robotic application programming interfaces (APIs). The experimental plan is then sent to and completed by automated instruments. In all, a human working with the system can design and run an experiment much more quickly, accurately, and efficiently than a human alone.

"Beyond the chemical synthesis tasks demonstrated by their system, Gomes and his team have successfully synthesized a sort of hyper-efficient lab partner," says

Q: How can we ensure that AI systems like Coscientist are not used for nefarious purposes?

The concept of extinction risk garners much attention, but it’s not considered an imminent threat based on our current understanding. Conversely, the risk posed by bioweapons is tangible and immediate.

Our focus should lean towards addressing immediate, tangible risks while preparing for future uncertainties.

That said, technological advancements always carry a spectrum of risk and benefit. With higher risks, we introduce more stringent controls or guardrails, along with appointed guardians to oversee these measures.

National Science Foundation (NSF) Chemistry Division Director David Berkowitz. "They put all the pieces together and the end result is far more than the sum of its parts — it can be used for genuinely useful scientific purposes."

In the Nature paper, the researchers demonstrated that Coscientist can plan the chemical synthesis of known compounds; search and navigate hardware documentation; use documentation to execute high-level commands in an automated lab called a cloud lab; control liquid handling instruments; complete scientific tasks that require the use of multiple hardware modules and diverse data sources; and solve optimization problems by analyzing previously collected data.

“Using LLMs will help us overcome one of the most significant barriers for using automated labs: the ability to code,” said Gomes. “If a scientist can interact with automated platforms in natural language, we open the field to many more people.”

This includes academic researchers who don’t have access to the advanced scientific research instrumentation typically only found at top-tier universities and institutions. A remotecontrolled automated lab, often called a cloud lab or self-driving lab, brings access to these scientists, democratizing science.

The CMU researchers partnered with Ben Kline from Emerald Cloud Lab (ECL), a Carnegie Mellon-alumni founded, remotely operated research facility that handles all aspects of daily lab work, to demonstrate that Coscientist can be used to execute experiments in an automated robotic lab.

"Professor Gomes and his team's ground-breaking work here has not only demonstrated the value of self-driving experimentation, but also pioneered a novel means of sharing the fruits of that work with the broader scientific community using cloud lab technology,” said Brian Frezza,

co-founder and co-CEO of ECL.

CMU, in partnership with ECL, will open the first cloud lab at a university in 2024. The Carnegie Mellon University Cloud Lab will give the university’s researchers and their collaborators access to more than 200 pieces of equipment.

Coscientist also, in effect, opens the “black box” of experimentation. The system follows and documents each step of the research, making the work fully traceable and reproducible.

"This work shows how two emerging tools in chemistry— AI and automation—can be integrated into an even more powerful tool," says Kathy Covert, director of the Centers for Chemical Innovation program at the U.S. National Science Foundation, which supported this work. "Systems like Coscientist will enable new approaches to rapidly improve how we synthesize new chemicals, and the datasets generated with those systems will be reliable, replicable, reproducible and re-usable by other chemists, magnifying their impact."

Safety concerns surrounding LLMs are paramount to Gomes. In the paper’s supporting information, Gomes’s team investigated the possibility that the AI could be coerced into making hazardous chemicals or controlled substances.

“I believe the positive things that AI-enabled science can do far outweigh the negatives. But we have a responsibility to acknowledge what could go wrong and provide solutions and fail-safes,” said Gomes.

“By ensuring ethical and responsible use of these powerful tools, we can continue to explore the vast potential of large language models in advancing scientific research while mitigating the risks associated with their misuse,” the authors wrote in the paper.

The simplistic solution of halting technological advancement is unfeasible. Instead, in this case, enhancing insight and transparency around large language models is crucial. Currently, companies are at the forefront of developing and deploying these technologies.

Creating preemptive protection mechanisms is essential before releasing models to the public. There’s a call for researchers from academia, government, and other sectors to have early access to cutting-edge models from major AI labs so that they can scrutinize them—a practice distinct from the internal red-teaming by these labs.

Q: Tell us more about the role of private industry?

Q: Considering risks, should we just stop the progression of technologies like Coscientist?

The implementation of safety measures requires robust support from the leading AI labs, as no government entity possesses the resources, knowledge, or speed for such undertakings. Given the significant yet complex benefits, responsible and vigilant oversight is imperative. Cloud labs exemplify part of the solution, offering controlled environments with extensive metadata where guardians can operate effectively. Open tracking serves as both a deterrent and a means to identify where additional safeguards are needed.

FROM THE DEAN

Artificial intelligence (AI) disrupted and irrevocably changed engineering education, and Carnegie Mellon College of Engineering is an undisputed leader in this transformation. We were integrating AI algorithms and machine learning into our curriculum long before others thought to do so. Our AI research is based on orchestrating the entire life-cycle of AI-enabled engineered systems, from design to sustainable deployment to operations, so that the functionality of a system is significantly improved.

Our magazine’s cover story reflects the salience of our approach. Assistant Chemical Engineering Professor Gabe Gomes and his students designed Coscientist, an intelligent system that can for the first time design, plan and execute a chemistry experiment on its own. Scientists can make this happen by using simple, plain language voice prompts. Coscientist is a powerful tool, and Gomes acknowledges that we must advocate for guardrails and guardians to ensure such systems are used ethically and responsibly, which aligns with the College’s values and teachings.

AI permeates the College and in the following pages, you’ll find articles that reveal how we are harnessing AI as well as other technologies to improve our lives. We have research underway that has resulted in a noninvasive method for detecting worsening brain injuries, a breakthrough with the potential to reshape neurocritical care. Then, of particular interest to industry is ongoing research into smart drones that

fly without GPS. Designing drones that can fly in underground tunnels without crashing into objects is no easy feat.

Another topic of great pride in the College is the new Scaife Hall that officially opened last November. The building’s well-equipped laboratories and modern classrooms create a welcoming environment that supports our proclivity for cutting-edge research. Faculty and students now have the space to share ideas and exploit technology as they seek solutions to engineering’s toughest challenges. New Scaife Hall is a shining example of our investment in future research endeavors and state-of the-art facilities for our students. I predict that Scaife Hall will foster great innovations and new entrepreneurs, and I consider myself lucky in that I always have exciting news to share about the College.

Whenever your travels bring you to campus, I look forward to sharing more about the incredible innovations happening across the College.

Sincerely,

William H. Sanders

Dr. William D. and Nancy W. Strecker Dean, College of Engineering

450-million-year-old organism finds new life in Softbotics RESEARCH

Humans have been walking the earth for roughly 300,000 years, an unfathomable period of time that in the grand scheme of Earth’s history is relatively recent. In fact, our time on earth represents only 0.007% of the planet’s history. Therefore, the modern-day animal kingdom that influences our understanding of evolution and inspires todays’ mechanical systems is just a fraction of all creatures that have existed through history.

To broaden our perspective of animal design and movement, researchers in Carnegie Mellon University’s Department of Mechanical Engineering, in collaboration with paleontologists from Spain and Poland, are introducing Paleobionics—a field aimed at using Softbotics, robotics with flexible electronics and soft materials, to

understand the biomechanical factors that drove evolution using extinct organisms.

“Our goal is to use Softbotics to bring biological systems back to life, in the sense that we can mimic them to understand how they operated,” said Phil LeDuc, professor of mechanical engineering.

Using computational simulations and soft robots, a research team led by LeDuc and Carmel Majidi have given pleurocystitid, a marine organism that existed nearly 450 million years ago, new life. Pleurocystitid, a member of the echinoderm class, which includes modern day starfish and sea urchins, were one of the first echinoderms capable of movement using a muscular stem. Despite the absence of a current-day analogue, pleurocystitids have been of interest

to paleontologists due to their pivotal role in echinoderm evolution.

“Softbotics is another approach to inform science using soft materials to construct flexible robot limbs and appendages. A lot of fundamental principles of biology and nature can only fully be explained if we look back at the evolutionary timeline of how animals evolved. We are building robot analogues to study how locomotion has changed,” explained Majidi, lead author and professor of mechanical engineering.

The team used fossil evidence to guide their design and a combination of 3D printed elements and polymers to mimic the flexible columnar structure of the moving appendage to build the robot. They demonstrated that pleurocystitids were likely able to move over the sea bottom with the aid of a stem that pushed the animal forward and determined that wide sweeping movements were likely the most effective motion. Increasing the length of the stem was also found to significantly increase the animals’ speed without forcing it to exert more energy.

“Researchers in the bio-inspired robotics community need to pick and choose important features worth adopting from

organisms over time,” said Richard Desatnik, Ph.D. candidate and co-first author.

“Essentially, we have to decide on good locomotion strategies to get our robots moving. For example, would a starfish robot really need to use 5 limbs for locomotion or can we find a better strategy?” added Zach Patterson, CMU alumnus and co-first author.

One of the biggest remaining questions about pleurocystitids are how the type of surface they lived on affected how they moved, be that sand or mud. Now that the team has demonstrated that they can use Softbotics to engineer extinct organisms, they hope to explore other animals, like the first organism that could travel from sea to land—something that can’t be studied in the same way using conventional robot hardware.

“Bringing a new life to something that existed nearly 500 million years ago is exciting in and of itself, but what really excites us about this breakthrough is how much we will be able to learn from it,” said LeDuc. “We aren’t just looking at fossils in the ground, we are trying to better understand life through working with amazing paleontologists.”

DETECTING BRAIN TSUNAMIS

A collaboration between researchers in engineering and medicine from Carnegie Mellon University, the University of Pittsburgh, and the University of Cincinnati has resulted in the first noninvasive method for detecting worsening brain injuries before they happen, a breakthrough with potential to reshape neurocritical care.

This approach is the subject of research by Pulkit Grover and collaborators that was published in Nature Communications Medicine.

Grover, a professor of electrical and computer engineering and the Neuroscience Institute at CMU, leads the For All Lab, a neuroengineering lab that works to develop novel diagnosis and treatment tools that span theoretical, computational, and hardware approaches, and are accessible to the broadest possible population. He explains that electroencephalography (EEG) is a routine procedure with no risks or side effects that is widely used in clinics.

“The patient is asked to sit in a chair or lie in a bed while a technician installs electrodes on their scalp to record brain signals,” said Grover. “Modern EEG is also portable, so people can walk around and carry on in their daily lives while signals are being recorded.”

This research marks the first time that EEG, or any noninvasive measurement technique, has been used to reliably detect a “brain tsunami,” a colloquial term for a spreading depolarization, which represents a wave of reduced activity that moves across the surface of the brain at a rate of a few millimeters per minute.

“Spreading depolarization is known to be predictive of worsening brain injuries, so detecting it enables use of corrective treatment,” Grover said. “There are no other known biomarkers that help you predict worsening brain injuries before they happen.”

The lead author on the work is Alireza Chamanzar, a postdoctoral research associate in electrical and computer engineering, who focused on the problem of detecting spreading depolarizations during his Ph.D. The team’s earlier work, on localizing neural silences, provided the foundations for this effort.

“Traumatic brain injury is a heterogeneous disease, and there is no objective diagnostic and treatment measure available to clinicians,” Chamanzar said. “Continuous monitoring of spreading depolarization, as a reliable biomarker and potential therapeutic target, leads to precision medicine and a reduction in secondary injury.”

Researchers have tried for decades to non-invasively detect spreading depolarization, recognizing its potential for improving patient care. One of the challenges is how to detect these traveling waves while dealing with a complex brain structure with folds. Having an interdisciplinary group work on this research was crucial so that each collaborator could use their respective expertise in engineering, medicine, or clinical research to make this lofty goal reachable.

The starting point was an algorithm called WAVEFRONT, developed originally in 2019 by Chamanzar, who was then a Ph.D. student in Grover’s lab. The first iteration of the model was trained with simulated data to measure electrode signals, but simulations are only so accurate. Real-world data was needed to take things to the next level.

WAVEFRONT was retrained using spreading depolarization data gathered from 12 patients via intracranial recording and electroencephalography, sourced from a study at the University of Cincinnati. WAVEFRONT validated this information with an impressive level of accuracy.

The team is actively working on improving the accuracy of their algorithm even further, especially to detect subtle signals from migraines and concussions. The end goal is to provide a new tool for clinical use that personalizes neurocritical care.

“For too long, patients have been grouped crudely into mild, moderate, and severe traumatic brain injury categories. This will help better stratify patients, and ultimately, provide better, more personalized care for this debilitating condition that affects millions every year,” said Grover.

The work was supported by the Chuck Noll Foundation for Brain Injury Research, the Center for Machine Learning and Health, and the National Science Foundation.

REVOLUTIONIZING ENERGY STORAGE WITH MXENES

We have demonstrated a way to utilize MXene that is reproducible and scalable. We believe this will impact energy storage devices.

With a slew of impressive properties, transition metal carbides, generally referred to as MXenes, are exciting nanomaterials being explored in the energy storage sector. MXenes are twodimensional materials that consist of flakes as thin as a few nanometers. Their outstanding mechanical strength, ultrahigh surface-to-volume ratio, and superior electrochemical stability make them promising candidates as supercapacitors—that is, as long as they can be arranged in 3D architectures where there is a sufficient volume of nanomaterials and their large surfaces are available for reactions.

During processing, MXenes tend to restack, compromising accessibility and impeding the performance of individual flakes, thereby diminishing some of their significant advantages. To circumvent this obstacle, Rahul Panat and Burak Ozdoganlar, along with Ph.D. candidate Mert Arslanoglu, from the Mechanical Engineering Department at Carnegie Mellon University have developed an entirely new material system that arranges 2D MXene nanosheets into a 3D structure. This is accomplished by infiltrating MXene into a porous ceramic scaffold, or backbone. The ceramic backbone is fabricated using the freeze-casting technique, which produces open-pore structures with controlled pore dimensions and pore directionality.

“We are able to infiltrate MXene flakes dispersed in a solvent into a freeze-cast porous ceramic structure,” explained Panat, a professor of mechanical engineering. “As the system dries, the 2D MXene flakes uniformly coat the internal surfaces of the interconnected pores of the ceramic without losing any essential attributes.”

As described in their earlier publication, the solvent used in

their freeze-casting approach is a chemical called camphene, which produces tree-like dendritic structures when frozen. Other types of pore distributions can also be obtained by using different solvents.

To test the samples, the team constructed “sandwich-type” two-electrode supercapacitors and connected them to an LED light with an operating voltage of 2.5V. The supercapacitors successfully powered the light with higher power density and energy density values than previously obtained for any MXene-based supercapacitors.

“Not only have we demonstrated an exceptional way to utilize MXene, we’ve done so in a way that is reproducible and scalable,” said Ozdoganlar, also a professor of mechanical engineering. “Our new material system can be mass-manufactured at desired dimensions to be used in commercial devices. We believe this can have a tremendous impact on energy storage devices, and thus, on applications such as electric vehicles.”

With outstanding experimental results and electrical conductivity that can be finely tuned by controlling the MXene concentration and the porosity of the backbone, this material system, recently described in an article published in Advanced Materials, has far-reaching potential for batteries, fuel cells, decarbonization systems, and catalytic devices. We may even see a MXene supercapacitor power our electric vehicles one day.

“Our approach can be applied to other nano-scale materials, like graphene, and the backbone can be built from materials beyond ceramics, including polymers and metals,” Panat said. “This structure could enable a wide range of emerging and novel technology applications.”

BROADENING MANUFACTURING WITH MICRO-ICE PRINTING

The Manufacturing Futures Institute encourages researchers to explore how advanced ice printing methods can expand manufacturing.

When Carnegie Mellon’s Manufacturing Futures Institute (MFI) issues its call for proposals each fall, it makes clear its objectives for offering seed funding to CMU faculty whose research is aligned with MFI’s mission to inspire, engineer, and lead technological and workforce advances for agile, intelligent, efficient, resilient, and sustainable manufacturing.

They are interested in funding research that falls into one of MFIs strategic priority areas; plays to CMU’s strengths in fields, such as AI and machine learning, coupled with advanced manufacturing technologies, such as robotics and additive manufacturing; benefits from convergent research and expertise from different disciplines; and shows potential for seeding the future of advanced manufacturing with new ideas that fuel innovation and attract future investments.

“We receive so many great proposals across a wide variety of advanced manufacturing topics that deciding which to fund is difficult. We focus

on the ones that propose innovative interdisciplinary approaches that can truly advance the state of the art. Industrial relevance is important as well, given our interest in technology transition,” said Sandra DeVincent Wolf, executive director of the MFI.

She and Gary Fedder, faculty director of MFI, reserve the right to ask for modifications to a proposal when timelines are too aggressive, funding amounts are not fully justified, or the project is missing a partner with vital skills and expertise.

That was the case when faculty members Phil LeDuc, Burak Ozdoganlar, and Charlie Ren requested funding for the next phase of their ice printing research. The team developed a novel approach to 3D print ice structures that are small enough to create vasculature in artificial tissue and other open features inside a fabricated part by creating sacrificial templates that later form the conduits and voids. It has proven to be a promising method. Still, it relied upon a lot of trial and error, as many automated approaches do

initially, in their development to achieve the specific intended geometry.

Creating more precise and reproducible geometries necessitates developing a feedback control approach, which would intelligently adjust parameters during printing to achieve the desired intricate geometries.

Since LeDuc and Ozdoganlar published their findings, which was featured on the cover of the Advanced Science journal in July 2022, they have received positive feedback at conference presentations, as well as support that has encouraged them to explore how their ice printing method can be further developed.

The initial biomedical engineering applications were clear and showed direct potential for use in personalized medicine. Still, the method also has the potential to revolutionize manufacturing by enabling the automated production of tiny internal 3D microchannels for many applications in various fields ranging from medicine to soft robotics. Micromanufacturing technology and

advanced approaches are fundamental to numerous critical applications across various sectors, including scientific research, healthcare, national security, and space exploration, as well as the digital transformation in industry that is at the heart of MFI’s manufacturing mission.

According to LeDuc and Ozdoganlar, the experimental and computational methods needed to build models to control the ice printing process are challenging. The microscale size and the fast nature of the thermal-fluidic processes and phase-changes (solid, liquid, gas) processes further exacerbate the challenges.

“We know that if we can build a sensor feedback system that can make adjustments in real-time, it will be very powerful for these approaches.” said LeDuc.

Wolf knew just the person who could help them develop the sensor hardware and computer vision algorithms needed to create a feedback control approach to produce the degree of control needed for printing

the microscale ice structures.

Lu Li, a project scientist at Carnegie Mellon’s Robotics Institute, had already brought his artificial intelligence and robotics expertise to several other MFI-funded projects. And according to LeDuc, Li has already made some great contributions to this new project.

With Li’s help, the team has reduced the time required for image segmentation from two to three seconds to 50 microseconds. Image segmentation is a computer vision task in machine learning that involves dividing an image into multiple segments or regions based on certain criteria so that the image can be changed into something that is more meaningful and easier to analyze.

“This is so typical of how we work at CMU. First, the ability to collaborate with someone like Lu Li, who has the skills we needed to move this forward, has been fantastic. And even though the MFI funding is not a huge amount of money, it has allowed us to keep moving forward with an idea that has huge potential,” said LeDuc.

ROBOTIC TESTBED ACTIVATES INNOVATION

Since the celebrated opening of Mill 19 four years ago, Carnegie Mellon’s Manufacturing Futures Institute (MFI) has been steadily outfitting the advanced manufacturing facility with state-of-the-art equipment to support its expanding research efforts and fulfill its mission to inspire, engineer, and lead technological and workforce advances for agile, intelligent, efficient, resilient, and sustainable manufacturing.

The latest addition to the facility is a flexible robotic testbed. Four six-axis industrial robot arms equipped with multiple sensors surround a worktable in the second-floor mezzanine area of MFI’s Mill 19 Building A location. Ceilingmounted programmable light curtains provide safety by sensing when humans get too close to the robots, and the system then immediately disables their motion.

Designed to initially test and advance robotic assembly and disassembly using Lego® bricks, the testbed is also capable of being used to study other robotic functions in manufacturing, such as material handling, quality inspection,

and more complex manipulation to thread small fasteners, to route electrical wiring, and even to assemble food items into snack packs.

Two autonomous mobile robots (AMRs) will be used to deliver Lego® bricks to and from the robot arms and will also be used in research aimed at training robots to move other materials and supplies efficiently and safely throughout a facility.

“We’ve built this testbed to accommodate both current and future research,” said Gary Fedder, MFI faculty director, who added that, “It is flexible enough to support current research efforts, as well as speculative projects that will be undertaken to compete for new federal awards and other future sources of funding.”

Like the additive manufacturing facilities and equipment at Mill 19 where students and faculty schedule time to conduct research on metals 3D printing, users will also need to reserve time to work with the robotic testbed. There are several

Lego® bricks are used to test a variety of robotic functions.

research projects underway to advance the robots’ ability to assemble and disassemble parts, to build a sharable codebase, and to support associated factory digital twin research.

Lego® bricks were chosen as the building material because they are not only safe and fun to play with, but they can also be used for constructing complex geometries that researchers need to test and build data sets. They are also ideal for creating simple structures that can be used to demonstrate robotic and digital manufacturing capabilities to students and industry professionals who visit Mill 19.

Shobhit Aggarwal, who earned his master’s degree in integrated innovation for products and services from Carnegie Mellon earlier this year, was hired as an advanced manufacturing engineer by MFI to oversee the testbed. In addition to coordinating its activity, he will also be helping develop short courses to demonstrate to industry partners and visitors how robots can be employed in assembly/ disassembly functions—a project that is funded in part by southwestern Pennsylvania’s federal Build Back Better Regional Challenge initiative.

“The testbed is essentially a production floor equipped with cutting-edge safety systems, logistics, a manufacturing execution system (MES), and industrial robots, providing an almost real-life environment to research, test, and gather data to develop groundbreaking technologies for manufacturing futures,” explained Aggarwal.

Nearly all the research at Mill 19 is being conducted to develop advanced manufacturing technologies. In addition to additive manufacturing and robotics research, the individual equipment and project cells like the robotic testbed at Mill 19 act as subjects for digital twin technology research.

Digital twin technology used in manufacturing creates virtual, interoperable models of equipment, production lines, or entire factories that update to physical twin data in order to aid in product design and development, process

optimization, predictive maintenance, quality control, training and simulation, and real-time monitoring and control.

The digital twin of the testbed will generate real-time data related to the equipment performance and provide insights that can make the system more efficient, agile and smart. As one of multiple project cells throughout the Mill 19 facility, the testbed will eventually be incorporated into a broader digital twin of the entire facility.

A growing codebase of utilities and APIs (application programming interfaces) associated with the testbed will enable both MFI researchers and visitors to more rapidly use the robots and focus on innovation.

The open-source software dubbed “MFI Codebase” will encompass an entire collection of source code, files, documentation, assets, and other resources for the software that will be used with other equipment and project cells throughout the facility.

A large video monitor near the testbed site, that can be used for presentations, will eventually also serve as a note board that will display and curate data streams of all the equipment in the building.

These advanced technologies require the expertise of a wide range of researchers from the College of Engineering, as well as those working in computer science and other academic areas through the university. The testbed area is also outfitted with multiple work areas designed to accommodate the interdisciplinary teams who meet with one another and educate visitors at Mill 19.

“The remarkable combination of human intelligence, digital innovation and advanced manufacturing technology and equipment at Mill 19 ensures MFI’s ability to conduct groundbreaking research, educate manufacturing innovators, and work with partners to translate our discoveries into industry relevant applications,” said Gary Fedder.

ARPA-H FAST TRACKS DEVELOPMENT OF CANCER IMPLANT TECH

The Advanced Research Projects Agency for Health (ARPA-H) has awarded $45 million to rapidly develop sense-andrespond implant technology that could slash U.S. cancer-related deaths by more than 50%.

The award was given to a multiinstitutional team of researchers, including Carnegie Mellon University, to fast-track development and testing of a new approach to cancer treatment that aims to dramatically improve immunotherapy outcomes for patients with ovarian, pancreatic, and other difficult-to-treat cancers.

The team spans 20 different research labs. The project and team are named THOR, an acronym for “targeted hybrid oncotherapeutic regulation.” THOR’s implant, or “hybrid advanced molecular manufacturing regulator,” goes by the acronym HAMMR.

“Using the key advantages of bioelectronics, we are focused on tackling one of the grand challenges of our time, the intense burden associated with cancer treatment,” said CMU materials scientist and bioengineer Tzahi Cohen-Karni, who serves as co-investigator and bioelectronics lead on the ARPA-H project. “Our aim is to develop a minimally invasive procedure to deploy a smart device that will monitor the state of the cancer and adjust the immunotherapy accordingly.”

Co-PIs on the project hail from Rice University, MD Anderson, Georgia Institute of Technology, Stanford University, Northwestern University, the University of Houston, Johns Hopkins University, the Chicago-based startup CellTrans, and New York City-based Bruder Consulting and Venture Group.

The THOR cooperative agreement includes funding for a first-phase clinical trial of HAMMR for the treatment of recurrent ovarian cancer. The trial is slated to begin in the fourth year of THOR’s five-and-a-half-year project.

“The technology is broadly applicable for peritoneal cancers that affect the pancreas, liver, lungs, and other organs,” said Rice University bioengineer Omid Veiseh, principal investigator on the ARPA-H cooperative agreement. “This kind of ‘closed-loop therapy’ has been used for managing diabetes, where you have a glucose monitor that continuously talks to an insulin pump. But for cancer immunotherapy, it’s revolutionary.”

Additional researchers on this project include CMU faculty members Doug Weber and Rahul Panat

The “hybrid advanced molecular manufacturing regulator,” or HAMMR, a “closed-loop,” drug-producing implant smaller than an adult’s finger is being developed to treat ovarian, pancreatic, and other difficult-to-treat cancers. The implant, which is small enough to be implanted with minimally-invasive surgery, will be able to continuously monitor a patient’s cancer and adjust their immunotherapy dose in real time.

Source: Brandon Martin/Rice University

NEW CENTER TO INVESTIGATE QUANTUM COMPUTING

A National Science Foundation Industry-University

Cooperative Research Center (IUCRC) at Carnegie Mellon University will create an ecosystem that advances quantum computing and information technologies. The Center for Quantum Computing and Information Technologies (QCiT) will bring together an industry consortium, government laboratories, faculty, and students to apply quantum computing and information technologies to solve practical problems that cannot be addressed with classical systems. The work of the center will be supported by the NSF and the consortium.

Quantum computing is estimated to generate $700 billion in value by 2035, with industries such as life sciences and finance to benefit first. While quantum mechanics has been long established, quantum engineering is emerging, and expectations are high as it could revolutionize AI.

Efforts to form the center at CMU coalesced in 2022 when Elias Towe, professor of electrical and computer engineering and materials science and engineering, led a team that hosted a workshop that laid the groundwork for the future center. Participants from industry and national laboratories, and faculty attended. They recommended creating a center that would support projects whose efforts focused on constructing two testbeds: one in a quantum communication network and another for a quantum computing processor.

The computing testbed is envisioned to promote research in algorithms used for materials simulations, drug discovery, industrial process optimization, and quantum machine learning. The quantum network would focus on secure communications, interconnecting networks for distributed quantum computing processors, and a quantum sensor network for high resolution and sensitivity measurements. The center will explore mathematical models and algorithms that bridge the gap between quantum hardware and practical industry applications.

The NSF approved the team’s proposal in August 2023, and the center was founded. The role of IUCRCs is to connect early academic research to commercial readiness. By collaborating with industry and government stakeholders, the center will identify applications that advance quantum computing while seeking solutions for difficult real-world problems. As the quantum ecosystem develops, the need for a trained workforce is expected to rise, and the center will

prepare students to move into the evolving field.

Prior to the formation of the center, a grassroots group of faculty across CMU were working on various quantum topics. The new center unites these researchers who represent different areas of expertise and reorient their focus toward goals relevant to industry.

“We want people from different backgrounds to understand the layers of the quantum ecosystem,” said Towe. “We want people who build the software to understand what the hardware people do and vice-versa. They may not deeply understand everything, but they can appreciate what others do.”

Other quantum activities in Pittsburgh

The Pittsburgh Quantum Institute, founded in 2012, and headquartered at the University of Pittsburgh, is an organization that promotes research and education in quantum science and engineering. The group’s members come from the engineering, physics, chemistry, and computer science departments at the University of Pittsburgh, Duquesne University, and Carnegie Mellon University. Benjamin Hunt, associate professor of physics at Carnegie Mellon, is the institute’s co-director. The institute encourages interested faculty, postdocs, and students to pursue quantum activities in the city. The group is developing external partnerships to expand the region’s quantum community and its capabilities.

AI TO AVERT AVIATION COLLISIONS

Pingbo Tang is creating AI tools that will one day aid workers in critical roles like aircraft control to avert risk and safely manage a complex system of moving aircraft. In his newest research, Tang, an associate professor of civil and environmental engineering, lays the foundation for future innovations in AI by identifying past factors that resulted in the loss of separation and then training a machine to recognize the factors associated with this risk to predict how better routing could prevent these scenarios.

Loss of separation is when two aircraft become too close, putting both at higher risk of a costly or potentially dangerous collision. Tang’s algorithm uses data taken via cameras monitoring the facial features of air traffic controllers, tracking and logging changes that correspond with a loss or near loss of separation incident.

Tang is working toward realizing several AI techniques for supporting the predictive management of facilities, workspaces, and civil infrastructure, all of which require algorithms like this that can quickly and efficiently process large datasets while overcoming missing data problems. Transfer learning is a machine learning technique in which researchers train a machine on large amounts of data with stored knowledge from similar problems and existing expertise on the subject. Such approaches can adapt models trained in some scenarios with sufficient training samples to other scenarios where data is limited.

Reinforcement learning is a machine learning technique involving training on large historical datasets. However, the machine is provided with different stored knowledge. It must produce a solution based on the analyzed data and clearly defined performance indicators of operational processes (penalties and awards that guide the machine to find the best strategies). Given the increased demands on a machine this would require, Tang says reinforcement learning is less tangible in the immediate future than a technique like transfer learning, which is informed by human expertise.

A tool based on reinforcement learning would be able to analyze datasets of aircraft positioning info—like those Tang used to train his map-matching algorithm—predicting and informing controllers how to avoid any taxi routes that would create a risk for loss of separation between aircraft, purely based on positioning data.

A tool based on transfer learning may use similar modeling and analytics. However, it would also be informed by information tagged by an expert and informed by stored knowledge from similar problems, such as his work on safety in nuclear power systems. Aircraft control involves a spatial system, while nuclear power plants are mechanical systems, but Tang can still draw many useful parallels for algorithms.

Tang’s work is part of a larger NASA University Led Initiative (ULI), which aims to create a prognostics and safety architecture for large engineered systems like our national aviation system.

VALIDATING LARGE LANGUAGE MODELS

Large language models (LLMs) are a class of artificial intelligence (AI) models that are designed to understand and generate human language. These models have been applied to language translation tasks and content recommendations, the latter of which relies on analyzing a user’s interests and preferences.

George Amvrosiadis, an associate research professor of electrical and computer engineering, describes LLMs as resources with great industry potential while also recognizing that AI must continue to improve its sophistication.

“In essence, large language models serve as versatile tools that can automate and enhance various languagerelated tasks, making communication, information processing, and content creation more efficient and accessible,” said Amvrosiadis.

“They have already demonstrated their versatility in addressing diverse real-world applications. However, they also raise ethical and societal concerns, such as bias in language generation and the responsible use of AI in decision-making processes and content attribution.”

One of the biggest challenges in dealing with language models is validating the information they generate. Amvrosiadis and Virginia Smith, an assistant professor of machine learning in the School of Computer Science, coadvised former Ph.D. student Michael Kuchnik, who took the lead on a research project to design an automated tool to audit LLM responses. The team’s research paper, “Validating Large Language Models with ReLM,” won the Outstanding Paper Award at the MLSys Conference in May 2023.

Validation, in the context of AI, refers to making sure that the responses generated by a language model adhere to

certain legal, ethical, and quality standards, while also being appropriate, accurate, and fair for the given context.

Considering language models are trained using vast amounts of text data from the internet, as Amvrosiadis explained, validating the responses they generate is important not only for factual accuracy but also to watch for and remediate biases. LLMs can inherit and propagate biases present in their training data, leading to the possible generation of discriminatory or offensive content.

In order for a tool to be practical, it must be based on the prompt-answer format used to query LLMs. The validation tool should be able to constrain the responses provided by the language model to a predetermined format. For example, when trying to determine George Washington’s birthday, the format of acceptable answers would be: “George Washington was born on <month> <day> <year>.”

Language models are essentially graphs with more connections than the neurons in a human brain, Amvrosiadis noted. To that end, getting a birthday right doesn’t prevent a model from spreading misinformation about George Washington in general. “Our tool, ReLM, can issue many questions per second to increase coverage against misinformation.”

There are still many advancements that Amvrosiadis foresees as LLMs continue to develop. “We also need tools that will allow us to act once we detect that a model has biases. LLMs are expensive to retrain, so we would have to find a way to mitigate responses that are inappropriate.

“I am hoping that ReLM is only the beginning in devising these tools, because LLMs are poised to become part of our everyday lives and affect the way we interact with the world.”

BRINGING DONOR ORGANS BACK TO LIFE

For end-stage diseases, organ transplantation is the ultimate treatment; however, the organ supply consistently underwhelms the demand. Adding to the reality of a globally growing transplant list is the need for a reliable technology to accurately determine if organs are suitable for transplant and to increase the number of these suitable organs.

Carnegie Mellon University’s Ren Lab, University of Pittsburgh Medical Center (UPMC) and University of WisconsinMadison are teaming up to reshape current practice related to organ transplants, by introducing a novel platform to study protein dynamics within the donor graft. The platform can be used to discover novel biomarkers to evaluate organ quality after donation, but before transplantation, as well as to reveal new therapeutic targets to improve donor organ function.

Most if not all biological processes are mediated by protein molecules, and changes in production and turnover of proteins are important to note. Current methods to study protein dynamics using mass spectrometry are generally good at detecting the most abundant proteins. They fall short in capturing what happens to proteins of low abundance, such as newly produced proteins that carry critical information related to donor organ function and injury response.

“We’ve developed a new technology that we’re calling newly synthesized glycoprotein profiling, to enable the characterization of proteins being newly produced within a short timeframe of a few hours, to better understand what a donor organ is going through, and how

it correlates with the transplantation outcome,” explained Zihan Ling, first author of the work published in American Journal of Physiology and biomedical engineering Ph.D. student.

Ling continued, “This engineering approach aims to bring donor organs back to life, to tell their ongoing story in terms of pathogenic protein production. By focusing on newly produced proteins, we can illuminate key details about the cause of an injury, not just the consequence. Understanding how the organ is injured at the protein level will help us better assess organ quality, so we can predict if it will cause serious complications after transplantation.”

The group’s initial investigation was conducted through an animal research study and centered on identifying molecular signatures of warm ischemic injury in lungs. Findings are broadly

applicable, however, and could translate to similar procedures for the heart, kidney, liver, and other organs. Futurestate, the group plans to test their technology platform on human lungs.

“A big part of engineering research is making your process so robust, so reliable, that you can ask very detailed questions,” explained Charlie Ren, associate professor of biomedical engineering. “We’ve been able to execute a highly relevant protein discovery program to see what’s happening in the donor organs. We believe this work has the potential to lead to dramatic improvement in donor organ quality and transplantation outcomes and ultimately to help alleviate donor organ shortage.”

This research was funded in part by the National Institute of Health.

Quantifying that pond smell

On a Tuesday in late June, Coty Jen and three students took their new kayak out on Green Lick Lake, southeast of Pittsburgh. There, the Pennsylvania Department of Environmental Protection (DEP) trained them to collect water samples from harmful algal blooms, as part of a new partnership.

When DEP gets reports of harmful algal blooms (HABs), they need to determine if it is safe for people to be in the water, for swimming or other recreation. Jen, an assistant professor of chemical engineering, and her research group are interested in determining if emissions from the algae could also be unsafe. “We know it’s harmful to swim in and ingest these blooms. We don’t yet know the inhalation effects,” says Christine Troller, a Ph.D. student in the Jen Lab who is leading this project.

Some emissions from algae are known to participate in reactions in the atmosphere that produce carcinogens. Jen and Troller are researching the potential aerosolization of toxins and compounds that may potentially become toxic.

In other words, they’re quantifying the bad smell you might have noticed near a pond.

The peak season for harmful algal blooms in Pennsylvania is usually mid- to late-August through September. As the climate warms, HABs are occurring earlier, more often, and in new places, like rivers. Runoff from farms and yards and sewage overflows are also factors. HABs are no longer only a rural problem. The risk of human interaction with harmful algal blooms is increasing.

In Pennsylvania, harmful algal blooms are most commonly caused by cyanobacteria. Troller will be sampling water from Green Lick Lake throughout the season. The DEP is analyzing biological toxins, and the Jen Lab is conducting air analysis, to see if there is a change in the concentration of cyanobacteria and the gaseous toxins they produce.

Troller starts measuring emissions the same day she collects the two-gallon samples, to get as close as possible to measuring the emissions from the algae in their natural habitat.

“There is some uncertainty because anything biological gets upset when you take it out of its environment,” says Jen.

In their lab in Doherty Hall, Troller puts each sample on a tank connected to a mass spectrometer. The gas phase emissions coming from the tank flow directly into the mass spectrometer. Over the course of a week, Troller continues measuring the gaseous concentration of hundreds of different compounds that are emitted from the water sample. She is identifying the compounds, many of which are unknown, based solely on their molar mass.

She will share information about potentially hazardous compounds with a PA human toxicologist. The findings will help determine if they can set standards based on gas measurements for when it is unsafe to be on the shoreline near a harmful algal bloom.

The Jen Lab is also developing two new instruments that could collect air samples from the environments where harmful algal blooms are reported. The goal is to create smaller, battery-powered, and less expensive instruments. Instead of being limited to collecting air samples from shore or taking water samples back to the lab, researchers could take these new instruments out on the water in a kayak, to measure the air directly above harmful algal blooms.

The Jen Lab’s larger interest in understanding the atmospheric effects of emissions from algae relates to cloud formation. The connections between particles and clouds are a major uncertainty in climate science. Compounds emitted by algae help form seed particles for cloud droplet formation. “There’s very little experimental data that modelers use to predict cloud formation,” says Troller. “Our research will add to the accuracy of those predictions.”

ENHANCING AQUIFER SYSTEMS

Groundwater supplies in the United States have been on the decline since the 1950s as climate change advances and dependence on groundwater grows. One long-established method for boosting groundwater supplies is enhanced aquifer recharge (EAR), or the replenishing of groundwater reserves

U.S. groundwater supplies are declining as climate change advances and need for groundwater grows.

with water from other sources.

Besides shoring up water supplies, EAR has broader benefits that can include mitigating land subsidence and saltwater intrusion, and even restoring ecosystems. However, until recently, EAR research efforts have disproportionately focused

on improving water quantity and not quality. The problem with this approach is that when water is injected into aquifers, it often disturbs geochemical conditions and mobilizes toxic contaminants, such as dormant naturally occurring substances like arsenic and uranium, or introduce new contaminants, such as pesticides. These substances can leach from sediments and soil and compromise the water. Adding to the complexity is that different regional locations have dissimilar geologies and hydrologic events (i.e., snow melts or flooding).

“Once you start artificially putting water into aquifers, you can really meddle with the geochemistry in those systems and create conditions that promote the contamination of groundwater supplies. That’s one reason why there’s been some hesitation surrounding EAR,” says Sarah Fakhreddine, assistant professor of Civil and Environmental Engineering at Carnegie Mellon University. Yet, the need to ensure the security and supply of water underscores the push to understand the complexities associated with EAR so that we can apply it in safe, innovative ways.

Fakhreddine, who is a geochemist by training, is the lead investigator on a three-year, $1.8m grant from the U.S. Environmental Protection Agency (EPA) that will increase our knowledge of the potential mobilization of contaminants and their risks to water quality in U.S. aquifer systems.

“Enhanced aquifer recharge is the idea that when we have excess water, like during a wet season or flood, we can capture that water and store it in aquifers for times when we really need it. Our goal is to make sure that when EAR is implemented, we fully understand the opportunities and limitations. And one of the big challenges is that the geochemical processes that control water quality really depend on where and how you implement a project,” says Fakhreddine.

Factors to consider include the water source used for recharge, the area’s hydrogeology, current geochemistry of the aquifer, and how the aquifer has been historically managed.

Fakhreddine and her team will conduct a nationwide study assessing the geochemical compatibility of aquifers with nearby water sources, which can include excess surface water and treated wastewater. The team, comprised of geochemical and hydrology experts, will leverage publicly available datasets, modeling tools, and community input to create a framework to access risks and implement safe EAR projects nationwide. They will use their findings to develop guidelines for siting, designing, and operating EAR projects so that contaminant mobilization is limited. Outputs from the project will include a guidance document, technical report, and a web-map tool to help decision-makers, including water managers and regulators.

The researchers highlight that “there is currently no water quality guidance that spans regional settings, source waters and EAR types to support safe adoption of fit-for-purpose EAR.” Fakhreddine’s project could change that, and ultimately help provide water to communities affected by extreme heat and drought.

“This study is very much focused on managed aquifer recharge, but that is just one water management practice among many,” says Fakhreddine. “It’s important to remember that there is no silver bullet solution to the water challenges we face in the future. Resilience is going to come down to having a diverse portfolio of different management practices that water managers can use.”

In addition to CMU, other research institutions participating in the project include the University of California, Davis and the University of Texas at Austin, in partnership with the Ground Water Protection Council.

PREDICTING PORTABLE POWER

Researchers design a model that predicts a battery’s charge curve, which could impact the safety and reliability of electric vehicles.

Lithium-ion batteries are the go-to power source for many of our favorite devices like cell phones and laptops, and their presence will continue to expand as electric vehicles become the new standard, replacing gasolinepowered cars.

Using a portable power source requires top-notch safety and maintenance features. A team of researchers from Carnegie Mellon University and the University of Texas at Austin has designed a battery management system to run crucial diagnostics on battery health so that drivers are able to make informed decisions. The model looks at two key diagnostics, the state of charge and state of health. In the short term, looking at the state of charge lets drivers know if they have enough power to reach their destinations, while in the long-term, battery health data can determine whether it’s time to replace a battery based on its ability to hold a charge relative to when it was new.

Reeja Jayan explains that while battery management systems already exist in most electric vehicles, there are a few qualities that make this new model stand out from the rest.

“We had a database of around 11,000 experimentally collected charging curves for a particular battery cathode chemistry,” said Jayan, an associate professor of mechanical engineering. “We used them to train a machine learning model to predict complete charging curves using sparse data inputs.”

What’s unique and practical about this research is that this machine learning based battery management system takes minimal data—just five percent of a battery’s charge curve—and produces a charging prediction whose margin

of error is two percent. Data collection itself is much simpler, too, requiring only about 15 minutes of charge time to calculate the curve and determine battery health. This information can even be collected in increments, so even if charging were interrupted, it would not ruin the analysis that was in progress. With such an impressive accuracy rate, this model was used to make predictions on an entirely different cathode chemistry. An updated version will be less data-driven and instead incorporate physics components into a calculation of a battery’s charge curves. While the laboratory-collected datasets were useful for training the original machine learning model to make predictions, this source also had limitations because it didn’t help researchers pinpoint the specific factors that cause batteries to fail.

“The charge curves we used in the study were collected at a constant c-rate and at room temperature,” Jayan said, “but charge current and temperature vary a lot in real-world battery applications. Collecting and using real data as input for the machine learning models will be an important next step to improve the model.”

Using environmental factors to calculate a battery’s charge and eventually discharge curves would go beyond a level of complexity that humans could produce.

“Right now, we have a model that uses an unsupervised peak recognition algorithm to identify specific features in battery charging curves,” Jayan says. “Neural network models take things a step further by learning why there might be a change in the shape of the battery’s charging curve—if a particle breaks, for example. Such a correlation between the curve shapes and various battery degradation mechanisms will

be critical for prediction of both battery performance and safety in the future.”

Not only is the teams’ system both interpretive and predictive, it’s also transferable to batteries made of different cathode chemistries. While lithium cobalt oxide has been the gold standard for decades, it can also be pricey. More and more, manufacturers are producing batteries made of other materials, so it makes sense to have a system that’s adaptable to them. (The model was first tested on lithium nickel oxide cathodes and was then transferred successfully to lithium cobalt oxide.) Accurately assessing battery health is also important for allaying safety concerns when dealing with emerging battery materials because their longevity and endurance haven’t been studied so extensively, but they will likely be relied upon to accommodate an increasing number of vehicles.

One of the most useful resources for the future of this research area is data taken from electric vehicle batteries that are out on the road. Using real-world figures and complex neural networks will allow battery management systems to make charge and discharge predictions with increasing accuracy, which will create a ripple effect on how electric vehicles are maintained as they become more commonplace.

Jayan and her collaborators have applied for a patent for this research, and their paper, “Battery Charge Curve Prediction via Feature Extraction and Supervised Machine Learning,” details the model’s capabilities and was published in Advanced Science

Advanced Science, first published: 02 July 2023, DOI: (10.1002/advs.202301737)

A look at how the machine learning model predicts a battery’s charge curve in the short term and assesses its longterm charging capacity relative to when the battery was new.

BATTERY CHARGE CURVE PREDICTION VIA FEATURE EXTRACTION AND SUPERVISED MACHINE LEARNING

SMART DRONES THAT FLY WITHOUT GPS

Commercial drones that use GPS signals work well outside, but it is a different story when it comes to flying indoors or underground in tunnels. In confined spaces, drones hit moving objects.

Mechanical Engineering Professor Kenji Shimada’s research overcomes this limitation by outfitting drones with an eye and a brain. The eye is a camera that makes a three-dimensional map of the environment, finds the drone’s location on the map and then detects moving objects. The brain, which relies on powerful onboard computing and smart algorithms, predicts the movement of objects and adjusts the flight path to avoid collisions.

Shimada’s work has industry supporters who build tunnels for cars and trains. They want to fly drones to measure the 3-D shape of new tunnels being excavated so they can compare actual data to design data.

According to Shimada, his drone can fly in tunnels, avoid moving objects, and accurately measure 3-D space. “Our sponsors are interested in using our technology in actual construction operations, and we have been discussing how to make a practical version of our drone system,” he says.

Smart drones can fly in underground tunnels without hitting objects.

INSIDE THE COLLEGE

CyLab celebrates 20 years of cybersecurity leadership

CYLAB celebrated 20 years of collaborating on cuttingedge research and educating the next generation of security and privacy professionals at its annual Partners Conference.

The conference, which was held October 3-5, 2023 at Carnegie Mellon University, highlighted the latest research in security and privacy with an interactive forum between faculty, students, industry, and government.

The hybrid event included three student poster sessions and a forum featuring more than 40 faculty and student presentations that focused on seven topics, including:

· Generative AI and machine learning

· System and hardware security

IoT security and privacy

· Human factors in privacy and security

Software security

· Blockchain and crypto

· Privacy

The conference also included a special 20-Year Panel, featuring CyLab leaders from throughout its history, that shared personal stories and memories of how CyLab grew from an idea amongst a few CMU faculty members into a worldwide leader in providing expertise on security and privacy issues.

“Looking back, I don’t think [CyLab] could have been created at any other place except CMU,” said Pradeep Khosla, founding director of CyLab and chancellor, University of

California San Diego.

“I also think that the combination of actors was perfect, and it was the perfect time to do this.”

During the forum, Aaron Roth, the Henry Salvatori Professor of Computer Science and Cognitive Science at the University of Pennsylvania, was officially recognized as CyLab’s 2023 Distinguished Alumni Award winner. Roth presented his latest research and received the award from his former advisee, Steven Wu, now assistant professor in CMU’s School of Computer Science.

In addition to the forum, participants had the opportunity to tour CyLab’s Biometrics Lab, take part in demos from WiSE Lab and picoCTF, and network during the 20-Year Celebration Gala at the Heinz History Center.

CyLab partners who attended the event say the conference provides them with insight into the latest trends in cybersecurity and privacy and offers the opportunity to connect with colleagues and academics doing leading work in the field.

“One thing that I would say makes CyLab very different than a lot of the other research institutions is that most of the topics that they bring are the ones that we see successfully implemented in industry,” said Ilyas Iyoob, chief data scientist and global head of research at Kyndryl. “That gives us confidence to keep working here because we know whatever is built is going to see the light of day at the end.”

“It’s been great to not only meet the students and faculty but also meet the other partners and see that we're part of an elite group—partners who are committed to not only the growth of CMU and CyLab but also the growth of these organizations as a whole, and to bringing security to the forefront of what we do,” said Jose Romero-Mariona, associate director of engineering at Raytheon Technologies.

“As a government entity, we see certain problems that academia can overcome,” said Melissa Chua, head of capability development (Cyber AI) at the Singapore Defence Science and Technology Agency’s (DSTA’s) Cybersecurity Programme Centre. “And we see CyLab having certain interests in cybersecurity that are similar to what we are looking at. We are partnering with CMU to understand the research work and share the outcomes with the professors more intimately, so we are able to derive value when we bring it back to our own internal use cases.”

“I really like the interdisciplinary nature of CyLab just as part of its existence,” said Tim Vidas, principal engineer at Amazon Web Services. “I think that that really influences the types of solutions that they seek, and their generally out-of-box thinking that doesn't lean too hard on the traditional methods and procedures.”

“To me, if you are forward-thinking and you want to see the trends in the industry from a service perspective, this is the place to be because we have students who are working on things that we don't think about on a day-to-day basis, and you get that research being done here by very smart people,” said Max Wandera, director of Eaton’s Product Cybersecurity Center of Excellence.

CyLab’s partners include a wide variety of businesses and institutions, each united by a passion for creating a world in which technology can be trusted. To learn more about partnering with the Carnegie Mellon CyLab Security and Privacy Institute, contact Michael Lisanti, director of partnerships, at mlisanti@andrew.cmu. edu or 412.268.1870.

CYLAB ANNOUNCES NEW FUNDED PROPOSALS

CyLab’s Future Enterprise Security Initiative has announced its second round of funded proposals. Each project falls under one of the four FutureEnterprise@ CyLab key research thrusts:

AI-driven workflows to automate security management and datadriven decision-making to minimize the need for large human teams.

Collaborative capabilities for real-time global visibility for security decision-making.

Foundations for understanding cyber risk and dependencies in complex ecosystems and supply chains.

Least-privilege-by-design infrastructure, including trustworthy outsourcing, remote work/management, and deployable software-defined architectures.

This year, Generative AI and Large Language Models (LLMs) were added as a technology of interest in all four research thrusts.

Funding for the projects is made possible by sponsorships from Amazon Web Services, Cisco, Microsoft, Nokia Bell Labs, PNC, and the VMware University Research Fund. Sponsors actively worked with FutureEnterprise@CyLab Co-Directors Lujo Bauer and Vyas Sekar on proposal requests and reviews.

“The Future Enterprise Security Initiative brings lots of value to CyLab because we get to benefit from our sponsors’ technical expertise and their understanding of which problems they're struggling with most. We direct our research towards solving the problems that matter now,” said Bauer.

CYLAB FUNDED PROJECTS

AI-Driven Workflows

Conversational AI to Simplify Wireless

Enterprise Security

PI: Swarun Kumar, ECE

LLM Self-Defense Against Adversarial Attacks for Coding Tasks

PIs: Corina Pasareanu, CyLab and Limin Jia, ECE

Risk

Assessment

Combining Program Synthesis and LLMs to Identify Code-Injection Vulnerabilities in Node. js packages

PIs: Ruben Martins, Computer Science Department (CSD) and Limin Jia, ECE

ODO: Open Dependency Observatory for Software Dependencies

PIs: Yuvraj Agarwal, S3D and Rohan Padhye, S3D

Collaborative Capabilities

Adversarial Robustness and Unhardening Dynamics in Federated Learning

PI: Carlee Joe-Wong, Electrical and Computer Engineering (ECE)

Evaluating Large Language Models’ Privacy Risks with Privacy Attacks

PI: Steven Wu, Software and Societal Systems Department (S3D)

Minimizing User Access by Design

Beyond Zero Trust Architectures for Enterprise Security

PI: Virgil Gligor, ECE

Verus: Enabling Engineers to Develop Provably Secure and Performant Software

PI: Bryan Parno, ECE and CSD

Adaptive Deployment of SDN/NFV Network

Security Infrastructure with SyNAPSE

PI: Justine Sherry, CSD

Provable and Practical Defenses Against Spatial Algorithmic Complexity Attacks

PI: Justine Sherry,  CSD

Enhancing Security and Portability with Lightweight Sandboxing Using the WebAssembly Linux Interface

PIs: Ben L. Titzer, S3D and Anthony Rowe, ECE

Teams working with the AI Coach have shown:

• Quality of the project and design goes up significantly

• Improvements are optimized

• Unforeseen changes are responded to faster

• Team member learning and satisfaction goes up

Changing the way teams perform for better results.

Learn more at: engineering.cmu.edu/ai-coach.

CAGAN DISCUSSES AI-HUMAN TEAMING AT CONGRESSIONAL BRIEFING

On September 27, 2023, Jon Cagan presented at the AI and National STEM Workforce Development Needs congressional briefing organized by the American Society of Mechanical Engineers, the Institute of Electrical and Electronics Engineers, and the Senate AI Caucus.

Conversation at the event was focused on opportunities and challenges regarding AI and machine learning in the workforce, including technology development, upskill training, and education. The session discussed the importance of AI tools for workforce development and resources to advance U.S. competitiveness in AI-enabled technology applications across industries.

Cagan, professor and head of the Department of Mechanical Engineering at Carnegie Mellon University, talked specifically about AIhuman collaboration and the impact AI has on team dynamics.

“AI can be a tool, a partner, and a guide for engineering teams,” he explained. “For example, we’ve found that AI-guided teams substantially improve their problem-solving process, resulting in significantly better outcomes. They enable people to think and act more strategically. At the same time, AI has limitations, and engineers need to understand what AI is, how it works, and when it works well."

Cagan underscored that resources need to be allocated to both upskilling engineers and R&D in AI engineering. He noted the importance of AI-centric engineering degree and certificate programs like those newly offered by the College of Engineering.

“AI is already having an impact in engineering practice and is here to stay.”

Cagan was joined by representatives from the U.S. Department of Defense, Purdue University, and IBM.

NEW FACULTY BOOK

While most teams and their managers know where they are headed, innovation is different; the problem objective has been outlined, but the journey and the destination are full of unknowns and untrieds.

In Managing the Unmanageable, Jon Cagan and Peter Boatwright, CMU professors, offer 13 tips that can greatly improve the odds for success for any innovation team. Filled with eye-opening realworld examples, bolstered by groundbreaking research studies, and enlivened with illustrations by artist Kurt Hess, it’s a quick, fascinating read that any manager with a mandate to innovate will find irresistible— and essential.

FACULTY AND ALUMNUS INDUCTED INTO THE NAE

Bill Sanders, dean of the College of Engineering and a professor of electrical and computer engineering, was among the 124 new members who were inducted into the prestigious National Academy of Engineering (NAE) during their annual meeting on Sunday, October 1, 2023.

Before coming to CMU in January 2020, Sanders spent 25 years as a professor of electrical and computer engineering and computer science at the University of Illinois. His research interests included secure and dependable computing and security, as well as resiliency metrics and evaluation, with a focus on critical infrastructures. He has published more than 300 technical papers and has also directed work at the forefront of national efforts to make the U.S. power grid smart and resilient.

He was the founding director of the University of Illinois’ Information Trust Institute in 2004, growing its faculty to more than 100 and attracting $80 million in external research funding by 2011. Sanders served as director of the Coordinated Science Laboratory from 2010 - 2014 and was head of the university’s Department of Electrical and Computer Engineering from 2014 - 2018. He also co-founded the Advanced Digital Sciences Center in Singapore in 2009, which was Illinois’ first international research facility.

Sanders was joined by Elias Towe, professor of electrical and computer engineering and materials science and engineering, and Carnegie Mellon engineering alumnus, Anirudh Devgan (ECE ’91 ’93), who were also elected to the academy’s Class of 2023.

Prior to joining the faculty at Carnegie Mellon in 2001, Towe was a professor of electrical and computer engineering at the University of Virginia, and a program manager at the Defense

Anirudh Devgan

Elias Towe

Advanced Research Projects Agency. His research is in basic optical and quantum phenomena in materials for applications in novel photonic devices that enable a new generation of information processing systems for communication, computation, and sensing.

Anirudh Devgan is president and CEO of Cadence Design Systems, Inc. Devgan, who joined the Pittsburgh-based company in 2012, is widely recognized as one of the leading authorities in electronic design automation. He pioneered the application of massively parallel and distributed architectures to create several industry firsts and most impactful products in the areas of SPICE simulation, library characterization, place and route, static timing, and power and electromagnetics, among several others.

National Academy of Engineering members are elected by their peers, and membership in the NAE is one of the highest professional honors accorded an engineer.

Academy membership honors those who have made outstanding contributions to “engineering research, practice, or education, including, where appropriate, significant contributions to the engineering literature” and to “the pioneering of new and developing fields of technology, making major advancements in traditional fields of engineering, or developing/implementing innovative approaches to engineering education.”

ALBERT PRESTO NAMED DIRECTOR OF CAPS

Albert Presto, research professor in the Department of Mechanical Engineering, has been named the director of the Center for Atmospheric Particle Studies (CAPS) at Carnegie Mellon University.

Presto, who received his Ph.D. in chemical engineering from CMU, has held a research role within the university since 2007. Focusing on pollutant emissions from energy extraction and consumption, Presto is a highly accomplished researcher publishing influential work on secondary organic aerosol, low-cost sensors, and exposure assessments including an examination of the aftermath of the train derailment in East Palestine, Ohio last year.

CAPS is one of the world’s leading research centers focused on particulate matter and its effects on global climate and public health. The center not only provides policy-relevant research but actively participates in the evolution of environmental policy related to particulate matter.

Presto is looking forward to strengthening CAPS’ community, which includes seven faculty members and 25 doctoral students across the Departments of Civil and Environmental Engineering, Chemical Engineering, Engineering and Public Policy, Mechanical Engineering, and Chemistry.

“CAPS is the perfect example of a ground-up scientific endeavor. We exist because a group of people at CMU were interested in working together on science and air,” Presto explained. “I want to build off of our grass roots nature and bring the community closer together.”

Presto plans to foster the community with recurring professional and social events.

“The people in CAPS are innovative,” said Presto. “Encouraging strong working relationships with new ways for interaction will keep our momentum going.”

Peter Adams, head of the Department of Engineering and Public Policy, served as the director of CAPS for the last decade. Adams states that “CAPS is a great example of CMU’s culture of interdisciplinary and department-spanning research. CMU and CAPS have a rich history of air quality research that includes leading major national research efforts, highly influential publications, and providing science advice to Allegheny County, the Commonwealth of Pennsylvania, and US EPA. I know Albert will do a great job, and I very much look forward to continuing my participation in CAPS under his leadership.”

DOE SEEKS CLEAN, CHEAPER HYDROGEN

The U.S. Department of Energy (DOE) has announced $750 million in funding for 52 projects to dramatically reduce the cost of clean hydrogen and reinforce American leadership in the growing hydrogen industry.

Shawn Litster was a sub-awardee on five of the selected projects with a total funding of $50M and a combined budget of $2.5 million directed to his lab over the next three years. Litster’s project partners include Plug Power Inc., 3M Company, Stanford University, pH Matter, and Power to Hydrogen.

Litster’s lab is focused on translating new developments in materials including catalysts and polymers to highperformance fuel cells and electrolyzers through a framework that encompasses advanced fabrication, experimental diagnostics, ultra-high resolution 3D imaging, and multi-scale simulation.

“What excites me about this funding is that the research will advance critical technologies for tackling many of the hard-to-abate carbon emissions sectors, including heavy duty transportation,” said Litster, professor of mechanical engineering and head of the Electrochemical Decarbonization Lab. “Long haul trucks are very difficult to electrify with batteries because of the weight and recharging time of batteries, so hydrogen fuel cells are a promising alternative for electrification thanks to the high energy density of hydrogen and the rapid refueling time.”

While two projects focus on fuel cells, three of the funded projects are on water electrolysis for hydrogen production from low carbon electricity sources.

“Electrolyzers to produce low-carbon hydrogen are critical to address carbon emissions associated with producing fertilizer, steel, aviation fuels, and cement,” Litster explained.

The projects selected for awards that Litster’s lab is a project partner on include:

High Volume Fuel Cell Manufacturing, Stack Assembly, and Final Test

Project lead: Plug Power Inc., $30M

This project will enable additional domestic manufacturing capacity of 20,000 fuel cell stacks per year from a global leader in clean hydrogen technologies. The project will demonstrate an innovative expansion of their current manufacturing line.

Durable, High Specific Power OER Catalyst/Electrodes for PEM Water Electrolyzers

Project lead: 3M Company, $5M

This project will develop catalysts with low-precious metal loadings for proton exchange membrane electrolyzers. These developments will enable catalyst cost reduction while maintaining performance and durability.

Towards Scalable Manufacture of Low Iridium Loading Catalyst for Durable PEM Water Electrolyzers (PEM-WE)

Project lead: Stanford University, $3M

This project will use theory to guide development of nonprecious metal catalyst supports for proton exchange membrane electrolyzers that can enable cost reduction without sacrificing performance and durability.

U.S.-Based Advanced Catalyst Manufacturing

Project lead: pH Matter LLC, $7.2M

This project will build on catalyst technology advancements to scale-up materials for use in the heavy-duty transportation market. The project will improve the domestic manufacturing base for catalysts and demonstrate the capability to meet DOE manufacturing capacity targets.

Advanced Electrolysis Cell Components Designed for Assembly

Project lead: Power to Hydrogen LLC, $6.6M

This project will further develop and scale-up a proven advanced liquid alkaline electrolyzer cell design that meets DOE performance targets while utilizing low-cost components. Components will be designed for GW-scale manufacturing and assembly while scaling the cell size. Developments will be applicable to a broad set of alkaline electrolyzer designs, helping to establish a U.S.-based component supply chain.

ZHANG RECEIVES NSF CAREER AWARD

Xu Zhang was awarded a National Science Foundation (NSF) CAREER grant for his research. The NSF Faculty Early Career Development Program awards grants to “early-career faculty who have the potential to serve as academic role models in research and education and to lead advances in the mission of their department or organization.”

Zhang, an assistant professor of electrical and computer engineering, received his Ph.D. from the Massachusetts Institute of Technology (MIT) before joining CMU. He founded the Zhang Lab, whose research focuses include device fabrication and system-level applications of atomically thin 2D materials.

This prestigious five-year funding opportunity will allow Zhang to continue his work on a specific class of kirigami-actuated adapted metasurfaces whose dynamic tunability makes them ideal for use in biomedical imaging, drone-based sensing and imaging, and wearable augmented reality glasses.

Zhang wants to make optical devices that take inspiration from fish, reptiles, and mammals–all of whom have adaptable irises that can adjust to different light environments. More specifically, Zhang will take an interdisciplinary approach to “build kirigami-inspired actuators and heterogeneously integrate them with tunable metasurfaces based on emerging nanomaterials,” according to the research abstract.

This ongoing project stands as an opportunity to experiment with the newest metasurfaces for the next generation of scientists and engineers who work in Zhang’s lab.

TEACHERS TURNED RESEARCHERS

Two new researchers joined the Jen Lab last July. Unlike most visiting researchers, Joe Hayes and Bryan Seybert are high school chemistry teachers. Through a National Science Foundation (NSF) grant, they spent the month working on research and curriculum development.

Coty Jen developed the NSF proposal to help increase atmospheric chemistry and climate change literacy in the Pittsburgh area. Jen, an assistant professor of chemical engineering, noticed that kids don’t retain very much from one-time outreach workshops. She wants to embed atmospheric chemistry into their science curriculum so students build intuition for how the atmosphere works. She’s hoping to encourage more people from all backgrounds to pursue careers in atmospheric science to better understand our rapidly changing planet. This pilot program draws on the expertise that teachers have working within curriculum requirements and teaching diverse high school students

Hayes and Seybert, the first participants, worked alongside Dominic Casalnuovo, a chemical engineering Ph.D. student, to look at how atmospheric ions impact air quality. “These ions are naturally occurring, and we poorly understand what they do,” says Jen. Her lab is studying their interactions with various gaseous pollutants.

“This research is about how chemistry directly impacts climate and our weather patterns,” says Seybert. “Like we saw with the wildfires in Canada this summer, the things we put into our air can have a very big impact on the formation of clouds and the shifting of climate patterns. That’s the most exciting thing I’m taking out of this. We have a chance to do some really cool stuff with our students that is truly relevant to them.”

The public schools where Hayes and Seybert teach, like others, are in the process of revamping their science courses by 2025, in response to new Pennsylvania science standards. “It’s time for a monumental shift in the way we teach chemistry,” says Seybert. “It’s a detail-oriented science. We get hung up on the particulars and don’t focus on the big ideas in science literacy.”

As part of the NSF grant, Hayes and Seybert are using their summer research experience to create a twoto-three-week unit on atmospheric chemistry. They are designing it to be used a la carte, so that other teachers can select specific topics to fit their curricular needs.

“The beauty of our match-up is that our schools are so different,” says Hayes. “What works in one classroom might not work in another. Together, we can make sure that what we roll out will be successful everywhere and appeal to the most teachers.”

Hayes teaches at Pittsburgh Westinghouse Academy, in Pittsburgh’s Homewood neighborhood. It offers fewer resources than South Fayette High School, where Seybert teaches in a growing school district southwest of Pittsburgh.

Both teachers share a commitment to make chemistry approachable and interesting to every student and not just the highest achievers. “That’s where this curriculum benefits all students,” says Hayes. “It allows them to make connections from what they’ve learned in class to better understand what they hear in the news and other media.”

During the school year, Jen will help evaluate and adjust educational content as Hayes and Seybert bring their summer learning into their classrooms. She is applying for additional funding to continue working with them for three more years. “Based on my own teaching,” says Jen, “I know it can take a few years to smooth out how new curriculum material works in the classroom.” Jen also envisions developing an exchange, where the teachers and their students will do experiments in their high school labs and then visit Jen’s lab to conduct additional experiments.

The vision is to make project-based learning a sustained experience. “We want to sprinkle it throughout the whole year, layered over and over, to help students understand and remember,” says Hayes.

STUDENTS TURNED CAREER CONSULTANTS

It’s obvious that Leigh Mason loves her job as the Associate Director of Career Services at CMU Silicon Valley. And why wouldn’t she—she gets to help some of the most talented and well-prepared engineering students find internships and jobs in the epicenter of technology innovation.

But with more than 300 students, who are oftentimes seeking competitive opportunities with prestigious companies, there is a constant demand for the career and professional development services she provides.

Despite her endless energy, Mason realized after taking on the role, that she couldn’t do it all herself. So, in the fall of 2019, Mason and CMU-SV Assistant Dean of Student Affairs, Lauren Schachar, launched the Peer Career Consultants program.

“I didn’t know if our students would be willing to counsel their peers, but I decided to try to recruit them and find out if they were interested,” said Mason.

Turns out they were more than interested—many were eager to help fellow students, and each year more students apply to the popular program.

In January, five new peer career consultants were hired based on their own job search experience as well as the passion they showed for wanting to help others. This is the fifth year of the program.

Mason runs a two-day training for the new peer career consultants in how to coach students, review resumes, conduct mock interviews, assess employers, and networking in Silicon Valley. She also has the consultants shadow her, practice counseling one another, and she shadows each of them as they begin consulting other students.

Dipam Paul, who will earn his M.S. in Software Management degree at the end of this year, says he appreciates the emphasis Mason put on listening and being empathetic with job seekers.

“We probably spend more time listening than giving advice. It’s important to understand their specific challenges so we can help them come up with effective strategies,” said Paul.

He says the tech job market has become more difficult recently and, like any job search, it can be stressful and demoralizing. He believes that as peers, the student consultants are especially good at relating to their struggles.

“They know I have been in the same boat. They know I get it, so they can be open and frank with me,” said Paul.

Shivani Ghuge is another new peer career consultant this year. She says that peer career consultants on campus who helped her in the past were excellent. She wanted to give back.

She also hoped that by becoming a career consultant she would learn to be a better job seeker. The training she received as well as the dozens of students she has already counseled have taught her a lot.

“Thanks to Leigh, we really learned to be careful in how we speak to students. She taught us how to put forth our ideas constructively,” said Ghuge.

In more formal sessions, Ghuge has reviewed resumes and helped students tell a better story by using stronger action words and highlighting relevant experience. When she conducts mock interviews, she prepares by studying the company a student plans to interview with.

Ghuge, who will also earn her M.S. in Software Management this year, also regularly attends campus events with alumni, entrepreneurs, and speakers from local companies.

“I am amazed at the network in Silicon Valley,” she says.

But networking is more than just an important job search skill that consultants teach. Each year, the group builds their own network and forms close bonds with one another.

“We’ve really become more of a family,” says Mason who is happy that there is now a tradition to hold an event each semester with both current peer career consultants and the alumni who have come through the program.

Mechanical Engineering Ph.D. students Michael Bennington (left) and Ravesh Sukhnandan (right) working with an adaptable soft grasping robot.

Innovation and creative ideas often blaze through boundaries.

Bending the rules, breaking conventions, pushing through collaboration, and pulling us forward. They bring change, create bursts of opportunity and elevate the world.

Big ideas drive the College of Engineering

Assistant Teaching Professor Brandon Bodily taught the new course to 12 enthusiastic students, who applied what they were learning to the semester-long team-based project assignment to engineer a product or service to improve the biking experience in Pittsburgh.

“Innovation really is a process that can be taught and learned,” said Bodily, who spent 20 years working in product development.

He says the broad scope of the assignment is ideal for first year engineering students in any major because they can think about solutions through whichever engineering discipline they’re most interested in.

“Civil engineers might want to consider the roadways, while electrical and computer engineers want to create devices,” said Bodily, whose enthusiastic and open-minded approach fostered the creative mindset the students needed to apply to the project.

The students responded to both the subject matter and Bodily’s flexible approach.

“Entrepreneurship has always been a goal of mine, and this new approach to literally integrate innovation and design was very valuable,” said Samy Penmasani, who added that she had already planned to pick up a major in entrepreneurship had she not found this program.

“But this is an even cooler version!”

Leo Hoplamazian agreed. He explained that “Applying engineering skills in a real-world context to solve problems is exactly what I want to do with my engineering degree.”

The four student teams developed ingenious and distinct ideas as they progressed through the innovation framework lessons that are the hallmark of much of the teaching at the III.

They followed a process that began by developing a problem statement to clarify what challenge their ultimate product or service could solve.

Next, they conducted both market

research and consumer interviews in order to frame a set of key findings and insights. They identified a product opportunity gap, to show how their new product or service could fulfill an unmet need. And they sought to better understand their potential customer by creating a buyer persona, a semifictional representation of their ideal customer, which included identifying what that customer most likely valued.

Using these values, the students were able to determine the product requirements that would need to be incorporated into their final concept, which led to the final phase to conceptualize their ideas. At the end of the semester, the teams presented their concepts to their classmates, instructors, and III Co-Founder, Director, and the Allan D. Shocker Professor of Marketing and New Product Development, Peter Boatwright, who had this to say to the class.

“I really appreciate the enthusiasm you showed. You obviously put a lot of effort into this, but I see that you really enjoyed it too – that’s what I was thrilled to see – you were passionate about it!”

That enthusiasm obviously spread to other first year engineering students. The class size for next semester grew nearly six-fold to 70 students having enrolled for the course in the spring session.

The budding entrepreneurs came up with four ways to improve biking in Pittsburgh.

Walter Light combines features of several existing products into one remote controlled device that featured mirrors and sound indicators to alert bikers when cars or pedestrians were nearby, and lights and electronic turn signals to make drivers more aware of bikers. Unlike a competitive product, Tether, Walter Light did not include navigation because the team’s research revealed that their target customer was a commuter who had little need for wayfinding on their daily rides.

IREADY is a wide mirror, mounted in the center of a bike’s handlebars that is made of a flexible material that allows the biker to customize the mirror for an optimal view. It is equipped with lights and ultrasonic sensors that alerts a cyclist when a vehicle has entered their blind spot. The team’s product requirements included a durable, easy to install, adjustable design that could be used with any type of bike that created a feeling of safety, control, and confidence—values they prioritized after interviewing bikers.

Electro is a detachable electric motor that clips onto a bike to make climbing Pittsburgh’s steep hills easier while also allowing bikers to retain the experience of free riding, which was an important feature to the bikers the team interviewed. Electro forfeits the 70-mile range of an e-bike for a shorter 30-mile limit, which the team believed would be adequate for most commuters. The shorter range could also equate to higher speeds, and the smaller size motor would take less time to charge. The team estimated that the cost of Electro would be about half as much as an e-bike.

Bike Link Express sought to improve public transportation options for Pittsburgh bikers who are dissatisfied with the current PRT (Pittsburgh Regional Transit) system. Bikers reported that it can be stressful to mount their bikes to the front of the buses, which inconveniences the other bus riders. Their idea, which was modeled after an existing program in Vancouver, would provide a special fleet of buses designed to transport bikers and their bikes. The buses would have wider doors and lowering systems that would make it easy to board bikes, which could then be easily secured with a simpler locking mechanism.

STUDENT NEWS

Integrating the Built, Natural, and Information Environments

REIMAGINING CRAIG STREET

CRAIG Street, which connects the main thoroughfares of Forbes and Fifth Avenues in Pittsburgh’s Oakland neighborhood, appears to serve the needs of the local university communities. But when students in Carnegie Mellon’s Civil and Environmental Engineering Junior Projects course worked with local professionals to consider ways to improve the busy two-way street, a somewhat different picture emerged. Their assignment was to develop an improvement plan for the congested four-block span that is lined with shops,

restaurants, and office buildings. Last fall, a class outing to observe how the street functioned, revealed numerous opportunities to transform Craig Street into a great college town street.

The steady car traffic on the narrow two-way street provides little room for cyclists. Parking, which lines both sides of the street, is at a premium. The sidewalks, typically crowded with pedestrians, is too narrow to comfortably accommodate outdoor dining and has too little greenery. The entire first floor of a large building that is owned by CMU is vacant. And perhaps

the most immediate need for change the students identified was the busy bus stop at the corner of Forbes and Craig, which is dominated by a row of large, dirty trash cans.

Joining the class that day were professional mentors who had volunteered to work with the students. Everyone was assigned to one of ten teams, each representing a different group of stakeholders to consider in developing their improvement plans.

Todd Wilson, who earned his bachelor’s degree at CMU in civil and environmental engineering and

engineering and public policy in 2006, was working with both the Traffic Engineering and the Roadway Safety stakeholder groups. He is a Senior Project Manager with GAI Consultants, who specializes in traffic-related civil engineering for transportation projects.

“This project is remarkably similar to what we do as professionals,” said Wilson.

This marks the fourth time that Wilson participated in the class. His experience as a student taking the course inspired both his chosen profession and his enthusiasm for supporting the current College of Engineering class project.

Two of Wilson’s colleagues from GAI Consultants also served as mentors. Keith Vasas worked with the Business Advocacy stakeholder group, and James Yost, mentored the Urban Planning group.

Karen Brooks, CMU’s campus bike consultant and Corey Harper, an assistant professor in civil and environmental engineering advised students in the Cycling stakeholder group.

Brooks serves on the board of Bike Pittsburgh and belongs to the League of American Bicyclists. This organization, which was founded to make bicycling safer and easier as a means of transportation and recreation, also rates the bike friendliness of college campuses, through its Bicycle Friendly University program.

“CMU is a silver awardee,” explained Brooks, who says that classes like this, which incorporate cycling into the curriculum is one of the factors the organization considers in making its awards.

Todd Reidboard, who mentored the Developer group, hosted the students at Bakery Square. He is the president of Walnut Capital, the developer of the trendy shopping and dining center, which would inspire the idea for a dining hall on Craig Street that the students dubbed Fusian Square.

Craig Toocheck, a Senior Planner at Pittsburgh Regional Transit advised the

Transit Agency group. He appreciated that the students conducted both qualitative and quantitative analyses.

Nizar El Daher mentored the Green Infrastructure group. He earned his Ph.D. at CMU’s School of Architecture and is the Construction Plans Examiner for the City of Pittsburgh. Pittsburgh City Councilwoman, Erika Strassburger, advised the Political/Policy group. And Jen Beck from CMU was the mentor for the Student stakeholder group.

The students and mentors met several times throughout the semester to gain a better understanding of the various Craig Street stakeholders’ needs and interests. They also talked to pedestrians, cyclists, and business owners on Craig Street.

Joe Moore, the assistant teaching professor, who has taught the course for four years was particularly impressed with the Traffic Engineering group, which independently organized their own traffic study to record the number and type of vehicles using the street.

After having served in the stakeholder groups, each student was then assigned to one of five teams that created plans for making Craig Street a safer, more beautiful, and welcoming destination, particularly for the college students.

Several groups favored changing

the roadway to one-way traffic and a single lane of parking so that dedicated bike lanes could be added. Elevated pervious concrete street crossings could improve stormwater management and make the street safer by clearly indicating crosswalks for both pedestrians and drivers according to the student plans.

Student presenters touted the benefits of enhanced lighting, signage, and wider sidewalks that could accommodate more green spaces and outdoor dining options. And thanks to their visit to Bakery Square, most of the student teams suggested that a food market or dining court be opened on the vacant first floor of the CMU building and several groups added a rooftop bar to the upgrade plan.

Their plans were backed by research the students conducted as well as input from their advisors, who guided the students through city ordinances and regulations, zoning considerations, commercial development issues, and National Association of City Transportation Official (NACTO) standards.

But perhaps the single most appealing idea that all five groups backed was moving the trash can bus stop from its existing location to a new space down on Forbes Avenue.

Students in Carnegie Mellon’s Civil and Environmental Engineering Junior Projects course.

EPP STUDENTS EARN AWARD AT NASA COMPETITION

For a group of Carnegie Mellon University researchers, hydrogen is the future.

Engineering and Public Policy (EPP) graduate students Jon Gordon, Jaih Hunter-Hill, Anna Cobb, and Xiaohan Wu, and CMU Design student Dorothy Li presented their research, “The Role of Hydrogen in Aviation Decarbonization," at NASA's Gateways to Blue Skies: Clean Aviation Energy Competition, winning the Best Presentation award.

The competition, which was sponsored by NASA’s Aeronautics Research Mission Directorate’s University Innovation Project, “is an initiative to engage college students

in researching climate-friendly technologies and applications that will establish a zero emissions future for aviation.”

The theme of last year’s contest asked students to conceptualize the life cycle of a clean energy option, from source to flight. Presentations were judged on feasibility, viability, and environmental impact.

Gordon said aviation is one of the hardest transportation sectors to decarbonize, but he feels there is real potential for hydrogen as a fuel source.

“We spent time looking into other alternative fuels that we thought could be interesting, but we ended up taking the approach that if we want our work to really mean something, we have to choose what we feel is the most realistic option,” Cobb said.

Moving forward, Gordon said their hope is that NASA, the Federal Aviation Administration, and the Department of Energy (DOE) will seriously consider their research and take specific action as a result.

“I hope our research for this NASA competition could provide valuable insights for decision-makers in the aviation

Award-winning team - top row: Anna Cobb (l.), Jaih Hunter-Hill (r.). Bottom row: Dorothy Li (l.), Jon Gordon (r.). Not shown: Xiaohan Wu.

industry,” Wu said. “Moreover, I also hope this research could bring a heated discussion on hydrogen's role in transportation decarbonization in the academic world.” The team summarized their work into a policy brief that they plan to share with the hydrogen and aviation community.

Gordon said the team feels that aviation is not discussed nearly enough in conversations about hydrogen end-users. He explains that the DOE is investing billions of dollars into Hydrogen Hubs—hydrogen production facilities that will be placed in eight to 10 areas across the country. He said if aviation is not part of the conversation as a hydrogen offtaker while these hubs are being created, it will delay hydrogen’s role in aviation decarbonization by years.

“If we act now, we can decarbonize aviation much more,” Gordon said. “How fast hydrogen is able to penetrate the market of aviation is dependent on the next few years of planning.”

Cobb explained the difficulties in decarbonizing aviation include the slow turnover rate for planes (30+ years), fueling infrastructure, and range.

“Being limited in range and fueling infrastructure introduces difficulties with flight scheduling; certain aircrafts can only fly to certain airports,” she said.

Gordon said they anticipate three types of energy sources to replace jet fuel: batteries, hydrogen, and sustainable aviation fuel. Batteries will be used for short-range flights while hydrogen can be used for longer travel times. While airlines have the goal of having net-zero emissions by 2050, Gordon noted they can’t reach that goal by using sustainable aviation fuel because it still produces emissions, albeit less than traditional jet fuel.

“We think it’s extra important for hydrogen to be part of the solution because it can actually be zero emission,” he said.

The team took a four-pronged approach to their research that included a literature review of more than 70 sources, interviews with 13 stakeholders, market share analysis, and a multi-objective optimization model. The goal of the

optimization model is to analyze trade-offs between their two objectives: minimizing emissions and minimizing costs.

Peter Zhang, assistant professor of operations research in the Heinz College and team advisor, said he was impressed by how independent and motivated the team was.

“The most impressive part is the amount of details and perspectives they bring into every meeting—a strong indicator of their enthusiasm and work ethic,” Zhang said. “In a good project like this, advisors have enough ‘ingredients’ to work with and can very quickly help them prioritize.”

The team also created and defined a supply chain system to illustrate how hydrogen will be transported from hubs to airports.

“With our optimization model, we're able to choose how we want to make hydrogen and, probably more importantly, how we want to move that hydrogen in terms of what's going to be the most cost effective versus emissions intensive,” Cobb said.

Hunter-Hill said one of the challenges in their research was coming up with realistic ways to achieve a future that does not yet exist.

“We had to come up with best-case assumptions for the world that could be. I think one of the challenges with that is so much of the analysis that we found is based upon very casespecific studies,” Hunter-Hill said. “We wanted to try to account for different outcomes in a world that we can't predict.”

Hunter-Hill said the creation of the supply chain process is one of the things that makes their research stand out.

“We’re trying to integrate multiple sources of information to create a future that doesn't exist but is realistic to potentially exist,” he said.

Doing research in a team environment, Wu said, was inspirational. He explained they were able to split up the work based on everyone's strengths.

“I think the biggest success was team cooperation,” Wu said. “Every team meeting was very productive and inspirational. Everyone cared about this project very much and devoted 100%.”

Jonathan Shulgach, a mechanical engineering Ph.D. student has a proven passion for robotics. He is fascinated with human-technology interaction and how it might expand in the future. In particular, he is reimaging efficient, affordable medical care by bringing patients who use prosthetics closer to the engineers who design them.

“I definitely see myself in the future utilizing existent technologies, incorporating them into robotics, and getting them in the hands of people in the best way that benefits society,” said Shulgach.

About five years ago, Shulgach met Dylan Beam when they were working together at the Rehabilitation and Neural Engineering Lab in Pittsburgh. “We both had a passion for rehabilitation and restoration of human abilities,” Shulgach said. “We were also into 3D printing and prosthetics. I was always interested in how we can make prosthetics better, and Dylan always talked about how we can make what currently exists available to more people.”

When the COVID pandemic put many things on pause, Beam founded a nonprofit called the Accessible Prosthetics Initiative (API) with its mission of increasing access to prosthetic technologies and care on a national scale. Shulgach joined the pilot team of volunteers. “It was about more than getting affordable prosthetics for people that don’t have them. It was also about trying to bring clinicians and patients together for a nonprofit,” Shulgach said. “I haven’t seen that before.”

Part of making prosthetics more accessible involves

CURING CANCER IS NOT ENOUGH A NEW COURSE FOR PROSTHETICS CARE

listening to users firsthand. To do this, API coordinated with Amputee Coalition, a national advocacy group for people affected by limb loss. API attended their online meetings and learned about each member’s prosthetic uses, as well as the obstacles they face in obtaining and affording a prosthetic that fits well. Those conversations also helped API members understand some of the ergonomic design challenges that exist with common prosthetics, and sparked ideas to pass on to STEM students learning engineering design.

Today Shulgach has settled into an educational role at the API. An engineering student himself and a former coach for robotics teams in Baltimore, Shulgach uses his insight to help develop curricula, which have been used in summer STEM programs. Although Shulgach is eager to build his mechatronics career, he’s also invested in the education of fellow aspiring engineers—particularly those who will go into the medical field—which led him to an opportunity to design a graduate bioengineering course at the University of Pittsburgh.

“Something that I’ve seen in industry and education is that the best way to prepare the next workforce generation is to make sure that they have technical literacy, or core comprehension of a skill set to safely and responsibly do tasks,” Shulgach explained.

“But for prosthetics in particular, it wasn’t easy to see commercially available devices that were following the same rate of growth as collaborative open-source design resources. So, one of the things that we were thinking of is, ‘What kind of

When Colette Bilynsky’s brother was diagnosed with childhood cancer, the impact it had on her family inspired her to begin a career in cancer research, hoping to make a difference by discovering a cure. After completing her undergraduate degree at George Washington University, she decided to pursue a Ph.D. in biomedical engineering at Carnegie Mellon University and has since begun to explore health policy in addition to her cancer-related interests. She believes that the cure for cancer alone is not enough—it must also be accessible and affordable for all.

tools can we give college-level students that are soon to go into the workforce in areas involving prosthetics and orthotics?”

It turns out that Goeran Fiedler, an associate professor in the University of Pittsburgh’s Department of Rehabilitation Science and Technology, had similar concerns. API and Fiedler collaborated to construct a master’s course, “User-Centered Design of Limb Prostheses,” which debuted during the Spring 2023 semester. The first half of the semester focused on training, while the second half consisted of projects where students applied core design principles and communication skills to create custom prosthetics for patients.

With Fiedler as the official course instructor, Shulgach and another API member, Matthew Shaw, were guest lecturers that taught a unit covering 3D printing technologies and 3D design and modeling. They emphasized the importance of choosing the right materials and methods to create a useful tool based on the specific needs of the patient. The course was designed to put students in a big-picture mindset and help them understand all the parts involved in connecting patients to resources.

“We wanted to give students the understanding they need to talk to prosthetists in the industry, and then, from a highlevel perspective, make decisions that bring patients and clinicians or prosthetists and designers together, to include everyone in the conversation and not make it a blind step-bystep-by-step process.”

Bilynsky is conducting her Ph.D. thesis research with Elizabeth Wayne, an assistant professor of biomedical and chemical engineering, who studies a type of cell in the immune system called a macrophage. Wayne’s lab seeks to understand how macrophages and monocytes can be used as therapeutic and diagnostic tools to treat cancer. Bilynsky joined the lab as it was being established, and she has had a lot of control over the direction of her research.

“I came in knowing I wanted to do cancer research, and it worked out well because Professor Wayne didn’t have anybody doing that in her lab at the time,” she says. “It gave me a lot of room to figure out exactly what I wanted to do and also gain a lot of problem-solving capabilities.”

Bilynsky has taken her work one step further as she explores the realms of health policy and science communication. She has completed two fellowships with the Jewish Healthcare Foundation in Pittsburgh, which have

allowed her to work with professionals from other fields such as law and occupational therapy. Her current project is focused on implementing a national patient safety board and proposes improvements to electronic health records for both physicians and patients.

Bilynsky is also involved in science writing. Not only does she write cancer articles for the blog “OncoBites,” but she is also completing a scoping review of colorectal cancer nanoparticle literature. Through these initiatives, she aims to help people both with and without scientific backgrounds understand cancer research and health policy.

“I’ve gotten to a point in my research where I know a lot about the field and have seen incredible solutions for treating cancer, but it’s all inaccessible to patients for a variety of reasons,” said Bilynsky. “This very much pushed me towards the health policy side as a way to do a lot of good in the world.”

PRESENTING NEW SCAIFE HALL

The new building more than doubles the size of the original Scaife Hall, expanding the footprint on Frew Street, forming an engineering and maker quad with Hamerschlag and ANSYS Halls, and creating a new entrance to campus.

Last November, Carnegie Mellon University celebrated the opening of the new Alan Magee Scaife Hall, which brings energy and vibrancy to the heart of the College of Engineering. The building’s unique design features shared laboratories that will support large, multidisciplinary teams—allowing researchers from across the university to work together in the same space. By increasing collaboration across disciplines, new Scaife Hall will propel emerging research in fields such as intelligent systems, Softbotics, and drones.

The next time you are on campus, take a few minutes to visit new Scaife Hall.

ALUM FORGES SUCCESS PUTTING PEOPLE FIRST

Growing up in a mill town south of Pittsburgh, Bruce Smith (MechE ’85) was all too familiar with the impact factory closures could have on a community.

During Smith’s senior year at Carnegie Mellon, he was connected with a group of unionized workers who were affected by the closure of an Oscar Meyer plant near Pittsburgh. The displaced workers were looking to start their own business and turned to the Mechanical Engineering Department for help.

“Seeing not only the financial impact, but the emotional toll unemployment took on these men and women made me recognize that my classmates and I might actually be equipped to help,” Smith says.

Because of that project and the passion of the workers who sought to bring back jobs for themselves and their community, Smith’s career mantra about putting people first was bolstered in an unforgettable way.

Today, Smith is the owner, chairman, and CEO of Detroit Manufacturing Systems (DMS), a certified Minority Business Enterprise that builds more than 1.5 million large-scale products per year and employs roughly 1,500 people, primarily from underserved areas.

Under Smith’s leadership, DMS has built an inclusive and empowering culture where employees are encouraged to be the boldest versions of themselves and have fun in the process.

“I’m enthusiastic about building up our employees personally and professionally and watching them light up in response,” said Smith.

DMS boasts an impressive internal promotion rate (one in three employees

have been promoted in the last ten years). The company also funds a program that pays tuition upfront so that employees can pursue further education in any discipline they chose to better themselves.

Smith emphasizes, “We really want to change the world one person at a time, which aligns with our core values of serving others, enriching lives, rising together, and giving back. These elements work in harmony to create an ecosystem where people work together and help one another.”

Smith credits CMU for lifting him up and ingraining the mindset that bringing the community together is pivotal to solving real-world problems.

To give back to CMU, Smith has made a gift to the Mechanical Engineering Department in support of a new initiative focused on developing activities within the undergraduate curriculum that center on societal challenges from energy and climate change to drought and disaster relief.

To the next generation of engineers, Smith says, “Have fun, make the world a better place, and help build up as many people as you can.”

Bruce Smith and his family

ALUM SEEKS PRACTICAL USES FOR QUANTUM COMPUTERS

When David E. Bernal Neira (ChemE ‘21) uses the word teleportation at work, he’s not chatting about science fiction. Bernal Neira is an associate scientist at NASA and the Universities Space Research Association (USRA), studying quantum computers with the hope of understanding them for practical uses.

“Anything that is small enough, cold enough, or isolated enough behaves according to certain laws of nature that are not the same as those we use at our scale,” says Bernal Neira. Behaviors at this scale are counterintuitive.

Quantum computing involves trying to use the phenomena that appear in quantum mechanics to perform computation more efficiently. “If we turn a problem that would take an impractical amount of time to solve into something that can be practically solved, then suddenly, we’re getting a lot of value, and we’re solving a lot of societal problems,” he says.

Bernal Neira works in the same group at NASA where he interned while a Ph.D. student at Carnegie Mellon University.

process systems engineering and quantum computing.

In addition to his internship with NASA, he also interned with ExxonMobil twice. The second time, he was part of a small task group trying to use quantum computing to optimize the maritime routing logistics of moving liquified natural gas around the world. “It’s a problem that gets complicated even for (non-quantum) computers, because the amount of choices that you can make grows exponentially,” says Bernal Neira. “Even computing one solution per second, it would take the age of the universe to get what the optimal solution looks like. That’s where quantum computing comes in.”

When his advisor, Ignacio Grossmann, was preparing to give a talk on the future of the process systems engineering field, Bernal Neira told him he thought it was quantum computing. A year later, Grossmann sent Bernal Neira a paper he had received from Sridhar Tayur, a professor

at Carnegie Mellon’s Tepper School of Business. Tayur was proposing to use quantum computers to solve the kind of optimization problems in which Grossmann is an expert. After reading Bernal Neira’s review of his draft, Tayur connected him to a quantum computing lab at NASA that needed an intern with experience in optimization.

Bernal Neira says he’ll never forget when Tayur asked him, “Do you want to work at NASA?” Those words represented an opportunity to pair

After completing his Ph.D., Bernal Neira was hired by Purdue University’s Davidson School of Chemical Engineering, alongside fellow chemical engineering alum Can Li (‘21). Bernal Neira started his faculty position last fall.

“Looking for optimal solutions is expensive. It’s time consuming. It’s complicated,” he says. “In the same way that Ignacio uses tools of mathematics and computer science to solve problems in chemical engineering, I am currently studying how we can use quantum computing, or how we can get inspired by quantum computing, in order to solve these problems more efficiently.”

David E. Bernal Neira

Alum invests to create change ALUMNI

FOR alumnus Greg Hamm (CEE ′73), investing in education is crucial.

Hamm and his wife, Wanda Ingmire, created the Hamm-Ingmire Climate Mitigation Endowed Research Fund, an asset that will provide the Carnegie Mellon University community resources on climate change mitigation, prediction, and adaptation. Hamm is also the second person to ever endow a CMU-Africa fellowship.

“My wife and I believe that if you want to invest in the future and make big changes, education is a really good way to do that,” Hamm said.

Hamm noted that while these causes—climate change and CMU-Africa—may seem separate, they both play a vital role in addressing global issues. You can’t solve one without solving the other, he believes.

“About 50% of the world lives in poverty by developed nations’ standards. If 50% of the world focuses on growth and 50% of the world focuses on climate change, neither problem will get solved. We have to control climate and have global development at the same time,” he said. “They’re both part of the same puzzle.”

Hamm has been doing research on climate change since his doctoral thesis at Stanford University in 1982 and felt the CMU College of Engineering was the best place to address climate issues.

“That has made this a 45-year-interest of mine,” he reflects. “I think Carnegie Mellon University is a fairly unique place to invest. It’s a school that gets a lot out of its investment and gets a lot out of its students.”

Hamm said the commitment of CMU students to hard work and of faculty to teaching fundamentals

continues to stick out to him, and he notes that the lessons he learned as a student have stood the test of time. He recalled being an undergraduate student at CMU and being surrounded by other talented peers, all committed to doing hard work.

As an example, Hamm remembers introductory computing courses not focused on computer language syntax but focused instead on in-depth concepts.

“At Carnegie Mellon, it was looking at, ‘What can we teach that’s going to last and be fundamental?’ rather than, ‘What are we going to teach that will allow students to solve this problem set,’” he said.

In terms of his investment in CMU-Africa, Hamm strongly believes the region has the largest gap between resources and talent, and the fellowship is specifically geared toward providing resources for African women.

“I’m convinced there’s plenty of talent in Africa, but the resources that allow that talent to really accomplish something are very limited in comparison to the developed nations,” he said.

Hamm said learning about CMU-Africa was one of the factors that encouraged him to start giving back to his alma mater.

“In the later stages of my career, I started doing a little more investing in the future,” he said. ‘The CMUAfrica program really put it over the top. I was very interested in Africa and that really connected with me.”

Hamm’s passion for his work continues throughout changes in his career, as seen by Ingmire.

“It's very pleasant to see this person I love and respect in different roles,” she said.

ADDITIVE MANUFACTURING: ALUM’S QUEST TO ADVANCE PART QUALIFICATION

Alumnus Luke Scime (l.), who was the first student to apply machine learning to additive manufacturing at Carnegie Mellon, is now a staff scientist at Oak Ridge National Laboratory who is developing process monitoring software to qualify additive manufactured parts at scale.

When Luke Scime was a mechanical engineering Ph.D. student at Carnegie Mellon in 2018, he took on a small side project to apply computer vision techniques he was learning in a computer science course to the additive manufacturing (AM) research he was conducting in Jack Beuth’s new lab.

Those early experiments to apply machine learning to some of CMU’s first metal 3D printers has grown in scope and scale and now encompasses a nationwide effort to develop software that is being used at 15 U.S. government labs and more than a dozen companies and universities.

Scime is now a staff scientist at Oak Ridge National Laboratory (ORNL). He and a team from the Manufacturing Demonstration Facility at ORNL have been developing Peregrine, a scalable in-situ process monitoring software stack that uses artificial intelligence to generate data to support additive manufactured part qualification. Like the peregrine falcon that the software is named for, their software is fast, agile, and adaptable.

“Luke was the first student of mine to apply machine learning to AM process monitoring applications. He made a big impact on future research in our group,” said Beuth, a professor of mechanical engineering.

NEED ABILITY TO QUANTIFY PARTS

Additive manufacturing offers numerous advantages, including the ability to create complex and intricate geometries that can’t be produced using traditional manufacturing methods. It can reduce material waste and supports a wide range of materials, including various metals and alloys. AM also reduces lead times by eliminating the need for tooling and molds and can produce parts ondemand. But the ability to fully realize the many significant advantages of the growing range of AM applications depends upon the ability to qualify the finished parts.

Scime came to campus in November to meet with students and faculty and deliver a presentation about his work at the Manufacturing Demonstration Facility, a Department of Energy user facility focused on early-stage research and development to improve the energy and material efficiency, productivity, and competitiveness of American manufacturers.

He outlined a progression of how in-situ data can support AM part qualification. At the most basic level, the data and meta data are recorded but generally exhibit low quality and inconsistent formatting. The second level provides better quality data that is recorded more consistently and is available for use in artificial intelligence applications that can detect anomalies. At the next level, in-situ data are spatially registered with ex-situ characterization data and enable physics-based modeling that can predict flaws and material properties.

The final aspirational application of in-situ data will generate reliable property predictions that allow AI algorithms to automatically iterate both the part design and the manufacturing process steps.

The challenge is developing a standard solution that can be applied to the many different types of printers, printer manufacturers, users, use cases, and materials, as well as a wide range of methods to monitor, store, and analyze the data that researchers need in order to build effective computer models.

Scime told his Carnegie Mellon audience that, “Many national labs are doing in-situ monitoring work, but it would not be efficient for every one of them to come up with their own data

management structure. If we can provide a flexible framework that they can all use, then their time and money can expand into some of the other important work that needs to be done.”

TODAY’S STATE OF THE ART

Currently, the Peregrine team is gathering data from nine laser powder bed fusion systems, four electron beam powder bed fusion systems, and six binder jet systems, as well as more than 10 different systems operated by their government and industry partners. They have loaded more than 2,500 builds into the software and are currently adding approximately 50 new builds each month.

The original scope of the project was smaller and its goal, which was similar to the project he worked on at Carnegie Mellon, was simply to use images to find anomalies. But the ORNL team knew they couldn’t create a solution for every individual printer, so the key research objectives are now to develop high-resolution, multi-modal, sensor-agnostic deep learning algorithms to identify in-situ and ex-situ data registration techniques and create common data formats and clean software interfaces that can be applied to a broad range of printers, users, and use cases.

The technology autonomously collects and analyzes layer-wise imaging data; provides remote monitoring and process intervention capabilities; tracks metadata and part information; produces advanced visualizations of both the underlying in-situ data and machine learning results; and enables identification of correlations between in-situ data and process parameters or ex-situ characterization data.

Relying on image data or sensing modalities that can be converted into image data, the code allows for various steps of pre-processing of the data, such as registration; distortion correction; annotation of training ground truth data; processing of data online or offline; creating a 3D rendering of the detected anomalies; and implementing corrective actions.

The goal for 2024 is to begin incorporating software features developed at other DOE laboratories back into the Peregrine code base.

“We are now working with more than 20 government, academic, and industry partners, who are contributing software features, providing unique datasets, and performing tests in production environments that will allow us to continue to grow Peregrine’s capabilities and generalizability,” explained Scime.

Beuth can’t help but feel a sense of pride in the work his former student is doing.

“I see the Peregrine software having significant impact on the AM community as a whole,” said Beuth, who reports that Carnegie Mellon’s Next Manufacturing Center is now working to acquire Peregrine.

“At that point Luke’s impact on CMU will have come full circle.”

Oak Ridge National Laboratory

IMPROVE

YOUR

CAREER PERFORMANCE

BRING NEW AI SKILLS TO YOUR ORGANIZATION

Advance your career by bringing the value of AI-enabled digital twins into your workplace.