Orb[i]s | AADRL
Thesis | 2018-2020
To our Families
Contents
Preface
12
Research
18
Initial Tests
36
Unit Development
62
System Aggregation
82
Structure Formation
124
Behavior
146
Collective Intelligence
176
Prototyping
198
Illumination Strategies
240
Bibliography
271
Orb[i]s | AADRL
Thesis | 2018-2020
Preface
Preface Introduction
“Without culture, and the relative freedom it implies, society, even when perfect, is but a jungle. This is why any authentic creation is a gift to the Future.� - Albert Camus
Culture A cultural space is not defined by a physical entity; rather it’s a metaphorical expression for communication and sociability. It is the communication that constructs meanings of various places. These spaces play a vital role in shaping a city, with the profound impact it has on its inhabitants.
City A city in the age of constant change and development is an ideal site of cultural diversity. With constant changes in the economy, development, job opportunities and growth, cities have a large influx of people from varied backgrounds, contributing to a wide spectrum of cultural expectations. Cities also tend to a large floating population that adds to the constant change. In order to promote and protect cultural interests, an adaptive system should be used in the creation of cultural spaces that can constantly address this latency and change.
Our vision Over the decades, as art has resonated culture, drawing an important link between humans and culture; we envision the creation of cultural spaces that synthesis art, technology and culture, encouraging use today as well as in the years to come.
Preface
Studio Brief
Constructing Agency: Future Culture 3.0 In order to address the Future of Culture, architecture must be explored through the lenses of mobility, transformability and self-structuring systems. Our work will traverse the path of adaptive forms that can reorganize and be autonomous to move beyond form. We envision the creation of a realm that has its own predispositions and will be choreographed by self-assembling and structuring systems that will curate spaces beyond conventional building typologies. The Spyropoulos studio explores how behavior-based design methods can be used to reconsider cultural projects for today, through the development of self-aware practices that see architecture as an infrastructure to address latency and change. This infrastructure would be intelligent and machine learned in order to support its adaptive attribute and cater to the environment it is set in. Our intent is to develop new relations in the ever evolving and adaptive human-machine ecology.
Orb[i]s | AADRL
Thesis | 2018-2020
14
Statement Thesis Statement
Orb[i]s is a Design – Research Project aimed at addressing the constant change in urban environments, with need for urban infrastructure that adapts to this constant change. Therefore, Orb[i]s is a proposal for a Prototypical System that is highly adaptive in nature, creating spatial infrastructure in the urban environment that will respond to the changing needs, numbers and development within the city, augmenting daily experiences and activities. Orb[i]s is a behavioural assembly that opens up the possibilities of a dynamic environment that is not limited to a building plan, rather is autonomous, adaptable, dynamic and self–assembling based on real-time data culminating in a constantly reconfiguring ecology. A sensory system encourages decision making in the system, making it self-aware. The self-assembling quality of the system arises from unit to unit communication, leading to higher orders of organizations and structures. The smallest scale of the system, being the unit itself, has the ability to be mobile, sense, and make decisions. Upon coming together with other units, the system demonstrates collective intelligence and plethora of changing behaviours, providing a higher level of flexibility and adaptability.
Preface
The highly adaptive nature of the system springs from the ability of a single unit’s mobility with the addition of a multitude of such units; As well as transform-ability at the unit level, with its ability to telescopically extend and connect to other units. These connecting arms bring about considerable transformation within an existing structure, without the need for more units. In addition to its adaptive, space-making ability, the system provides a layer of interaction with its users. Culminating from research of light and its various effects, illumination has been incorporated in the system to provide an added layer of experience. The ambience created by the play of light and shadow, though not physically yet brings about transformation in the space generated. In all, our ecology is envisioned to provide spatial and experiential infrastructure, augmenting and aiding urban change.
Orb[i]s | AADRL
Thesis | 2018-2020
16
Research
Introduction
London as its been from back in the day to today, has seen nothing but constant change. A city like London, has always been extremely diverse, fast paced and an ever changing metropolis with constant development and need for infrastructure. With such diversity and contrasting factors, catering to the varying desires of large masses is always a task. Also keeping in mind the drastic weather patterns which further adds to the ordeal successful of social and cultural engagement. The most popular trend and desire means of engagement in a city so diverse is through art and culture. Art based activities are always drawing masses and promoting engagement. Although, gaining access to such engagement is not always viable. Most of events like these are highly labor intense, specific to particular locations and time specific. What we need, is a means of catering to these ever changing needs. What we need is a system, that will not only adhere to these changing factors but cope with and enhance experiences through time. A system that will aid and enhance daily experiences whilst addressing the change and latency that surrounds us. Future of Culture.. Orb[i]s
Research
Orb[i]s | AADRL
Thesis | 2018-2020
20
Case Studies Self Organizing Systems Social Insects : Behavioural Design Strategies The design of the Orb[i]s system, is inspired by the behaviour of social insects. Just like a colony of social insects where individual activities do not need a supervisor, yet they achieve a high degree of organization. Self-organization emerges out of interactions among individuals with a simple set of behaviours. Important features of self-organization are the flexibility and the robust means of adaptation, allowing for efficient performance of a system. The path to problem solving in these systems aren’t predetermined, they are rather emergent and result from interactions amongst individuals and their environment. Patterns of self-organization are generally based on nearest neighbour interaction. In all, a self-organizing system is one which is dynamic, where structures appear at the global level of a system from interactions at lower level components. Based on these design strategies, our system will use a similar approach in enabling autonomous behaviour in each Orb[i]s. Using a Bottom-Up approach, the system will a culminate into a multi-agent system from simple unit-to-unit interaction.
Research
Kilobots : A thousand Robot Swarm Swarm Intelligence is the discipline that deals with natural and artificial systems composed of many individuals that coordinate using decentralized control and self-organization. Examples of systems studied by swarm intelligence are multi-robot systems and certain computer programs that are written to tackle optimization and data analysis problems. The Kilobot swarm is a thousand-robot (1024) swarm designed program and experiment with collective behaviours in large-scale autonomous swarms. Each robot is equipped with the basic capabilities required for an autonomous swarm robot (programmable controller, basic locomotion, and local communication), but is made with low-cost parts and is mostly assembled by an automated process. Additionally, the system design allows the user to easily scale and operate larger Kilobot collectives. The Kilobot swarm was used to investigate collective artificial intelligence (e.g. sync, collective transport, self-assembly) as well as to explore new theories that link minimal individual capabilities to achievable swarm behaviours. By using a combined theory-experiment approach, the aim was to develop new algorithmic insights into robustness, scalability, self-organization, and emergence in collectives of limited individuals.
Self Assembly Algorithm
Orb[i]s | AADRL
Thesis | 2018-2020
22
Case Studies Interactive Systems Light Pollination : Illumination - Response Systems Light Pollination is an interactive digital artwork commissioned by iGuzzini to celebrate the power of light as a vehicle for social innovation and was designed by London-based digital-arts studio UniversalAssemblyUnit. The installation seeks to spread the word about light, and in doing so, it explores the strong links between light and communication. Fiber optic, the primary material used to create the artwork, is a vehicle for light through which high-speed communication is facilitated. Thus, the art installation is both an expression and a prototype of this, albeit on a smaller scale. Rather than addressing a particular function, it imagines an alternative way of interacting with artificial light.
Research
Fluid Assembly Chair : Self Assembly Lab, MIT Fluid Assembly is part of a series of investigations by MIT’s Self-Assembly Lab looking at autonomous assembly in complex and uncontrolled environments (water, air, space etc). In this experiment a number of components are released into a tank of turbulent water. Each of the components is completely unique from one another and has a precise location in the final structure. The process was filmed over 7 hours, after which a full assembled, precise chair was created. The chair was selected to demonstrate differentiated structures as opposed to repetitive growth or self-similar structures. This experiment points towards an opportunity to self-assemble arbitrarily complex differentiated structures from furniture to components, electronics / devices or other unique structures. Once self-assembled, the structures can be removed, tested, used or disassembled and thrown back into the chamber.
Orb[i]s | AADRL
Thesis | 2018-2020
24
Research Colour and Lights The Subconscious Effect of Coloured Lighting Keeping with the latter theme of our vision, we wish to integrate illumination in our system. Illumination is envisioned to add a layer of experience and utility to the system as a whole. Light creates more than just visual effects (image, shape, intensity, perception, contrast, etc.); it also has biological and psychological effects that can impact the health and wellbeing of humans. Brightness, hue and saturation are the three main qualities of light in relation to colour. Brightness is the amount of light given off by a light source, usually expressed in lumens or lux. Some studies have shown that brighter light can intensify emotions, while low light doesn’t remove emotions, but keeps them steady. Hue is defined as a colour or shade. Colours created by artificial light can evoke different emotions and have effects on the human body. Saturation is the intensity of a colour. More saturated hues can have amplifying effects on emotions, while muted colours can dampen emotions. In art, saturation is defined on a scale from pure colour (100% [fully saturated]) to grey (0%). In lighting, a similar scale can apply. Colour subconsciously effects our everyday life. It is used strategically to elicit a specific feeling in people. While each colour has a separate meaning, they also each have a separate feeling associated with them, affecting our mood in everyday situations.
Research
IMPACT OF COLOURS GREEN: is known as the strength provider. In lighting green can be used to portray nature, growth, cool, money, health, envy, tranquility, harmony, calmness, fertility, safety and ambition. BLUE: is known as the bringer of peace. Blue can be used to portray trust, loyalty, cleanliness, air, sky, water, health, tranquility. PURPLE: can help reduce emotional and mental stress, it portrays royalty, power, creativity, mystery. RED: can be used to portray love, warm, intensity, passion, danger, leadership, courage and friendship. ORANGE: known as the source of creativity, orange stimulates the creative thought process and can help people come up with new ideas. YELLOW: can sometimes be beneficial in the treatment for depression. In lighting design, yellow can be used to portray happiness, laughter, cheer, warmth. WHITE: can be used to portray purity, innocence, cleanliness, sense of space, neutrality, safety, beginning, faith and coolness.
Orb[i]s | AADRL
Thesis | 2018-2020
26
Research Illumination Systems Smart Highway project : Daan Roosegaarde Glowing Lines charge during day-time and glow at night for several hours to create an iconic highway experience and increase safety. The project is part of the Smart Highway. Glowing Lines uses photo-luminescent paint to mark out the edges of the road, and is the first of five concepts to be realized from Dutch designer Daan Roosegaarde’s Smart Highway project – designed to make highways safer while saving money and energy. SMART HIGHWAYS are interactive and sustainable roads of tomorrow by Daan Roosegaarde and Heijmans Infrastructure. Its goal is to make smart roads by using light, energy and information that interact with the traffic situation. “Here the landscape becomes an experience of light and information,” said Studio Roosegaarde in a statement. “As a result this increases visibility and safety.” The lines are now installed along the N329 route in Oss for an initiative called Road of the Future. Three glowing green lines run along each side of the dual carriageway and illuminate every night. Roosegaarde described driving along the section of road at night as “going through a fairy tale”.
Glow lines illuminating by the night on the Smart Highways
Research
Parallels : Nonotak Studio Parallels is an audio visual installation by Noemi Schipfer and Takami Nakamoto, commissioned by STRP. Like in all their installations Parallels explores light as a material, but this time the space as a whole becomes their screen. The boundaries and notion of space, become abstract as the audience crosses the room, but in doing so, the audience also affects the space by breaking the light. This installation is strongly connected to the space in which it takes place; it lives within it. But as soon as the light hits the walls that define the space it reaches its limits and stops reproducing itself. The installation is also inspired by Anthony Mccall’s exploration of light and space.
The varied lighting effects providing a varying sense of space even within the same given space.
Orb[i]s | AADRL
Thesis | 2018-2020
28
Unit Geometry History of spheres
1. Cénotaph de Newton
Designed by Étienne-Louis Boullée in 1784, this unbuilt structure is considered a purely modern or futuristic invention. It is a reminder that they have been a source of visionary architectural inspiration and fascination for centuries. Given neoclassicism’s obsession with geometry and proportion and the Enlightenment’s preoccupation with technology and advancement, it is perhaps no surprise that some of the biggest and most ambitious spherical projects were conceived not in the 20th or 21st centuries but in the 18th. Chief among them was the stupendous plan for a French mausoleum to Sir Isaac Newton. Although Boullee’s stone and concrete monument was never built, at 150m high it would have overtaken the Pyramids to become (at the time) the tallest building in the world.
Research
2. Montreal Biosphère
Buckminster Fuller is the visionary architect-inventor who revolutionized the design of spherical buildings with his ground-breaking geodesic dome. Despite their name, geodesic domes are actually spheres whose surface is formed by a series of tessellated triangular elements. Fuller used this principle to develop a perfect form of sphere that was lightweight and potentially portable. He built several, including his own home, but his most famous is the pavilion he built in Montreal for Expo 67 – the 1967 world fair – which is now part of the Biosphère environment museum. An enclosed structure of steel and originally acrylic cells, the 76m wide and 62m high semi-submerged sphere inspired several imitators, including the famous 1982 Spaceship Earth (Epcot) sphere at the Walt Disney World Resort in Orlando, Florida. Orb[i]s | AADRL
Thesis | 2018-2020
30
Unit Geometry Advantages of a Sphere
As an object a sphere actually has the optimum surface-area-to-volume ratio: it’s very efficient. While looking at a bubble floating in the air, it’s a perfect sphere because that shape provides it with the optimum lightweight structure to support itself and withstand the forces around it. Imagine a raindrop hitting the earth – it’s forced to change its form and behaviour because the forces acting against it change. Dealing with those forces and maintaining the form presents a unique challenge. How a sphere is built is as critical to its feasibility as what it is built from. As James Law explains, new developments in modular construction could accrue efficiencies. “If a spherical building enjoys repeatable structural elements and easy connecting details, and if it can be based on a mass-produced and prefabricated set of structural components, it can be just as efficient and economical as a rectilinear form.”
Research
Mobility
Area
Packing ability
R
Optimum Surface Area to Volume Ratio
R
Perfectly Symmetrical
No defining edge
Orb[i]s | AADRL
Thesis | 2018-2020
32
Unit Geometry Packing Strategy
When it comes to packing 3D objects, spheres have one of the most efficient close packing possibilities amongst others shapes. Spheres are also easily re-configurable and re-adjust to connect with the neighboring units upon any disturbance in the system. This led us to choosing to study this geometry and its possibilities in detail, since our goal is to achieve growth in all six axes.
Several other geometries tested before spheres.
Millions of black plastic balls covering the Los Angeles Reservoir, in order to protect the region’s drinking water.
Research
Spheres provided the optimal packing ability in all conditions
96 million black plastic balls covering the Los Angeles Reservoir. Orb[i]s | AADRL
Thesis | 2018-2020
34
Initial Tests
Signal Transmission Seeding Patterns |
Variable start points of growth
Behavior Based on Neighbors |
Growth based on neighbor proximity
Initial Tests
Signal Transmission
Signal Transmission
Orb[i]s | AADRL
Thesis | 2018-2020
38
Component Iterations Iteration 1
Plan
Connection between 2 components
Initial Tests
Exploded Axonometric
Orb[i]s | AADRL
Thesis | 2018-2020
40
Component Iterations Iteration 1
Top view of Connection between 2 components
Front view of Connection between 2 components
Initial Tests
Top view of Connection between 3 components
Front view of Connection between 2 components Orb[i]s | AADRL
Thesis | 2018-2020
42
Component Iterations Iteration 1 | 2 Component transformation
1.
2.
3.
4.
Initial Tests
5.
6.
7.
8.
Orb[i]s | AADRL
Thesis | 2018-2020
44
Component Iterations Iteration 1 | 4 Component transformation and mobility
Transformation State 1
Transformation State 2
Initial Tests
Mobility Orb[i]s | AADRL
Thesis | 2018-2020
46
Component Iterations Iteration 2
Exploded Axonometric
Initial Tests
State Transformation
Orb[i]s | AADRL
Thesis | 2018-2020
48
Component Iterations Iteration 3 | Component Connections
SINGLE UNIT
FRONT VIEW
SIDE VIEW
NGLE UNIT
FRONT VIEW
SIDE VIEW
SINGLE UNIT Flexible joint
Flexible joint FRONT VIEW
Flexible joint
Rigid legs
TWO UNITS COMBINED
SIDE VIEW
Flexible joint
Rigid legs
Flexible joint
Flexible joint
Rigid legs
WO UNITS COMBINED
Ball joint detail
Ball joint detail
TWO UNITS COMBINED Ball joint
Ball joint detail
Ball joint
Ball joint
Initial Tests
Two units Two units Two units Two units
Four units Four units Four units Four units
Consolidated form Consolidated form Consolidated form Consolidated form
Expanded form Expanded form Expanded form Expanded form
PERSPECTIVE PERSPECTIVE PERSPECTIVE PERSPECTIVE
ELEVATION ELEVATION ELEVATION ELEVATION
PLAN PLAN PLAN PLAN
Single Unit Single Unit Single Unit Single Unit
Family of 1, 2 & 4 Unit connections Orb[i]s | AADRL
Thesis | 2018-2020
50
Component Iterations Iteration 3 | Tiling Patterns in Compressed State Two units Two units Two units
Four units Four units Four units
PERSPECTIVE PERSPECTIVE PERSPECTIVE PERSPECTIVE
ELEVATION ELEVATION ELEVATION ELEVATION
PLAN PLAN PLAN PLAN
Single Unit Single Unit Single Unit
4 Component Transformative States
Consolidated form Consolidated form Consolidated form
Expanded form Expanded form Expanded form
Initial Tests
Consolidated form Consolidated form Consolidated form
Four units Four units Four units
Expanded form Expanded form Expanded form
ELEVATION
PLAN
Two units Two units Two units
PERSPECTIVE
PERSPECTIVE PERSPECTIVE PERSPECTIVE
ELEVATION ELEVATION ELEVATION
PLAN PLAN PLAN
Single Unit Single Unit Single Unit
4 Component Transformative States Orb[i]s | AADRL
Thesis | 2018-2020
52
Component Iterations Iteration 3 | Aggregation with Different States
Closed State 1.
2.
3.
4.
Semi-Open State
Initial Tests
5.
6.
7.
8.
Orb[i]s | AADRL
Thesis | 2018-2020
54
Component Iterations Iteration 3 | Aggregation Patterns in Semi-Open State
1.
2.
3.
4.
Initial Tests
5.
6.
7.
8.
Orb[i]s | AADRL
Thesis | 2018-2020
56
Component Iterations
TILING PATTERNS OF COMPONENT IN COMPRESSED STATE
Iteration 3 | Tiling Patterns in Compressed State TILING PATTERNS OF COMPONENT IN COMPRESSED STATE
its
1 Unit
2 Units
3 Units
5 Units
6 Units
7 Units
2 Units
1 Unit
MPRESSED STATE
3 Units
3 Units
4 Units
4 Units
4 Units
5 Units
5 Units
6 Units
7 Units
7 Units
8 Units
8 Units
Initial Tests
Aggregation
Orb[i]s | AADRL
Thesis | 2018-2020
58
Component Iterations
Unit test 01
Unit test 02
Unit test 03
Initial Tests
Legs for x,y,z axis
Expansion capability
Expansion capability
Orb[i]s | AADRL
Thesis | 2018-2020
60
Unit Development
Component Iterations Iteration 4 | Transformation of a single component
Single Unit : Transformation SINGLE UNIT : Transfomation
SINGLE UNIT : Rotations
Two Units
Connection Transformation
Connections and transfomations
TWO UNITS
Three Units
THREE UNITS
Four Units
FOUR UNITS
SINGLE UNIT : Transfomation
Unit Development
Single Unit : Rotation SINGLE UNIT : Rotations
Eight Units
MULTIPLE UNITS
Multiple Units
Connections and transfomations
EIGHT UNITS
Orb[i]s | AADRL
Thesis | 2018-2020
64
Left + Down Rotation
Left + Down Rotation
Left + Down Rotation
Left + Down Rotation
Right + Right + Up Down Rotation Rotation
Right + Right + Up Down Rotation Rotation
Right + Right + Up Down Rotation Rotation
Right + Right + Up Down Rotation Rotation Left + Up Rotation
Left + Up Rotation
Left + Up Rotation
Left + Up Rotation
Rightward Rotation
Rightward Rotation
Rightward Rotation
Rightward Rotation
Leftward Rotation
Leftward Rotation
Leftward Rotation
Leftward Rotation
Downward Rotation
Downward Rotation
Downward Rotation
Downward Rotation
Upward Rotation
Upward Rotation
Upward Rotation
Upward Rotation
Component Connection Catalog
Connection between 2 Components
Upward Upward Downward Upward Downward Upward Leftward Downward Leftward Rightward Downward Leftward Rightward Left Leftward Rightward + UpLeftLeft Rightward + Up+Left Down Left + Up Right +Left Down Left ++ Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotatio Rota
Unit Development
rd ghtward ward Downward Leftward Rightward Left + Leftward Up Rightward Left Left+ +UpRightward Down Left Left Right + +Up Down +Left Left Up Right +Right +Up Down + +Up Left Down Right Right + Down + +Up Down Right Right++Up Down Right + Down on tation tion Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation Rotation
Orb[i]s | AADRL
Thesis | 2018-2020
66
Local Organizations Iteration 4 | 4 Component Connection & Mobility
1.
2.
3.
4.
Unit Development
5.
6.
7.
8.
This illustrates the unit-unit connection and the aggregation (tetrahedron pyramid) it results in. Orb[i]s | AADRL
Thesis | 2018-2020
68
Orthogonal & Organic Formations Regular Angle Connection
Unit Development
Variable Angle Connection
Orb[i]s | AADRL
Thesis | 2018-2020
70
Component Mechanism Rotation and Expansion
30
30
Unit Development
Outer Shell Rotating Mechanism (Outer Shell)
Extendable Leg (1) Extendable Leg (2) Magnet
Rotating Mechanism (Inner)
Fully Extended Leg
Orb[i]s | AADRL
Thesis | 2018-2020
72
Unit States Growth in X-Y plane Two types of units are employed in forming the proposed structures Active and Passive units. Each type of unit has a specific function and position in how the structure is formed. Active unit Passive unit
Growth with 8 components
Growth with 15 components
Growth with 22 components
Growth with 33 components
Unit Development
Growth in X-Y-Z plane The passive units are compacted with their legs retracted into the sphere which inhibits their ability to ascend vertically on the Z-axis. These units can move around, and connect to other units on the base layer but don’t have the capacity to illuminate. The active units possess full expansion capabilities of the legs and can light up when required to do so.
Passive Unit
Active Unit with illumination
Growth with 14 components
Growth with 50 components Orb[i]s | AADRL
Thesis | 2018-2020
74
Unit States
State -1 : Passive State
Expansion : 0 - no expansion Potential : Tangential Connections - Packing Ability
State 0 : Active Potential State R
Expansion of one unit: ‘R/2’ (half the radius of unit) Total Expansion : ‘R’ (radius of unit) Potential : Linear connections
State +1 : Active Executive State
Expansion of one unit: ‘R’ (radius of unit) Total Expansion : ‘D’ (Diameter of unit) Potential : Diagonal connections
Unit Development
Active State Abilities
Orb[i]s | AADRL
Thesis | 2018-2020
76
System Scales
Unit
Unit-Unit
Unit Development
Unit-Human
Unit-System
Orb[i]s | AADRL
Thesis | 2018-2020
78
2
System Aggregation
Deformation in a System Deformation in a 2D Mesh Case 1 | One Signal initiated on the periphery of the grid
System Aggregation
Case 2 | One Signal initiated inside the grid
Case 3 | One Signal initiated inside the grid
Orb[i]s | AADRL
Thesis | 2018-2020
84
Deformation in a System
Case 4 | One Signal initiated inside the grid
Case 5 | Two Signals initiated in the grid
System Aggregation
Deformation in a 3D Mesh Case 1 | One Signal initiated inside the System
Case 2 | One Signal initiated on the periphery of the System
Orb[i]s | AADRL
Thesis | 2018-2020
86
Component Aggregation Growth Triangular Aggregation
1 Component
Linear Growth with 10 components
Planar Growth with 10 components
Planar Growth with 200 components
Planar Growth with 50 components
System Aggregation
Tetrahedron Aggregation
1 Component
Linear Growth with 10 components
Planar Growth with 10 components
Planar Growth with 50 components
Planar Growth with 200 components Orb[i]s | AADRL
Thesis | 2018-2020
88
Component Aggregation Growth Natural Triangular Aggregation
Growth with 10 components
Growth with 50 components
Growth with 200 components
Growth with 1000 components
System Aggregation
Natural Tetrahedron Aggregation
Growth with 10 components
Growth with 50 components
Growth with 200 components
Growth with 200 components Orb[i]s | AADRL
Thesis | 2018-2020
90
Component Aggregation Growth Signal Transmission
This catalog is to study the trajectory of an activation signal as it passes through different aggregation growth patterns.
System Aggregation
Organised & Natural Aggregations
Orb[i]s | AADRL
Thesis | 2018-2020
92
Component Aggregation Growth
In order to achieve a larger System of detection of neighbours, their connections and reconfiguration to further achieve the most efficient configuration based on imbibed rule sets, an NK network could be developed as follows:
K=0 and N=5
K=1 : the better state of each node depends on the state of the node that depends on it and the state of the node it depends on.
System Aggregation
K increases : the better state of each node depends on the state of the nodes that receives its inputs and the states of the nodes that provide inputs. In this case N=5, K=2.
The wheel is more like an organisation where all the information is filtered through a central unit.
The circle topology indicates the interaction of units with their ‘k’ nearest neighbours
Orb[i]s | AADRL
Thesis | 2018-2020
94
Units Connection Possibilities Different connection options with a varying number of units is explored here.
-1 1.
2.
State
1
3.
-1 State 7.
8.
9.
13.
14.
15.
19.
20.
21.
1
System Aggregation
-1 4.
State
1
5.
6.
-1
State
10.
11.
12.
16.
17.
18.
Orb[i]s | AADRL
1
Thesis | 2018-2020
96
Unit Connection States
-1
State
1
-1
State
1
-1
State
1
-1
State
1
-1
State
1
2 units
3 units
3 units
4 units
4 units
System Aggregation
-1
State
1
-1
State
1
-1
State
1
-1
State
1
4 units
5 units
5 units
6 units
Orb[i]s | AADRL
Thesis | 2018-2020
98
Transformation in a Triangle Pattern In these tests of expansion in a triangle pattern system, we studied expansion as a line and as a surface to study what transformation could be achieved.
1.
2.
3.
1.
2.
3.
1.
2.
3.
1.
2.
3.
System Aggregation
1.
2.
3.
1.
2.
3.
1.
2.
3.
1.
2.
3.
Orb[i]s | AADRL
Thesis | 2018-2020
102
Transformation in a Pyramid Pattern In these tests, some parts of the system expand and some of them contract. We tested this to understand the transformative behaviour of the system.
System Aggregation
1.
2.
3.
1.
2.
3.
1.
2.
3.
Orb[i]s | AADRL
Thesis | 2018-2020
104
Transformation in a Square Pattern In these tests of expansion in a cubical system, we studied expansion as a line, as a surface and as a volume to study what transformation could be achieved.
1.
2.
3.
1.
2.
3.
1.
2.
3.
1.
2.
3.
System Aggregation
Expansion and Contraction in a Square Grid 256 units
1.
2.
3.
1.
2.
3.
1.
2.
3.
1.
2.
3.
Orb[i]s | AADRL
Thesis | 2018-2020
106
Transformation in a Square Pattern In a cubical system with a square connection pattern, transformation was studied by disconnecting some of the units and anchoring the bottom units. Once units went from State -1 to State 1, the transformation would vary based on the position of units that disconnect.
1.
2.
3.
1.
2.
3.
1.
2.
3.
System Aggregation
Expansion and Contraction in a Square Grid 1386 units
1.
2.
3.
4.
5.
6.
1.
2.
3.
4.
5.
6.
Orb[i]s | AADRL
Thesis | 2018-2020
108
Transformation in a Square Pattern In these tests, some parts of the cuboidal system expand and some of them contract. We tested this to understand the transformative behaviour of the system and how it could give us usable spaces.
1.
2.
3.
1.
2.
3.
1.
2.
3.
1.
2.
3.
System Aggregation
1.
2.
3.
1.
2.
3.
1.
2.
3.
1.
2.
3.
Orb[i]s | AADRL
Thesis | 2018-2020
110
Transformation in a Square Pattern The units expand and contract based on linear signals generated within the system.
Square Pattern State -1
Triangle Pattern State -1
Pyramid Pattern State -1
State 0
State 1
State 0
State 1
State 0
State 1
System Aggregation
Structural Analysis Cantilevering Unit
2x2x2 grid
6 Stable Units 1 Cantilevered Unit
2x2x3 grid
8 Stable Units 2 Cantilevered Unit
2x2x2 grid
12 Stable Units 2 Cantilevered Unit
6x6x5 Grid | 145 units
26 Units
34 Units
31 Units
30 Units
24 Units
Orb[i]s | AADRL
Thesis | 2018-2020
112
Arch Formation In these tests, by studying the different connection patterns, we studied if the transformation could create an arch. This was only successful in a pyramid system.
Square Pattern State -1
Triangle Pattern State -1
Pyramid Pattern State -1
State 0
State 1
State 0
State 1
State 0
State 1
114
System Aggregation
Arch Formation in a Pyramidal Structure 200 units
0.96m
55% Expansion
60% Expansion
State 0
1.44m
70% Expansion
65% Expansion
1.8m
1.6m
75% Expansion
2.2m
85% Expansion
1.2m
80% Expansion
2.4m
90% Expansion
2.6m
95% Expansion
2.8m
100% Expansion State 1
Orb[i]s | AADRL
Thesis | 2018-2020
Grid Formation The units sense and connect in a gridded manner along the axes of the legs. Once the first layer is completely filled, the units start climbing over to the next layer. Next, the units start optimizing space occupation and try to occupy maximum number of available spaces in the grid
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
System Aggregation
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
Orb[i]s | AADRL
Thesis | 2018-2020
116
Triangle and Pyramid Formation The units connect to form a triangle in one layer with the legs fully extended. Then with the aid of a helper unit,another unit climbs to the next layer. The unit then rotates and extends it legs to connect and form a pyramid.
1.
2.
3.
4.
5.
6.
7.
8.
9.
Triangle Formation
System Aggregation
10.
11.
12.
13.
14.
15.
16.
17.
18.
Pyramid Formation Orb[i]s | AADRL
Thesis | 2018-2020
118
2
Structure Formation
Structure Formation The Units assemble and disassemble to create different typologies of spaces dependent on user needs and patterns.
1.
2.
3.
4.
5.
6.
Structure Formation
1.
2.
3.
4.
5.
6.
7. Orb[i]s | AADRL
Thesis | 2018-2020
128
Structure Formation
1.
2.
3.
4.
5.
6.
7.
Structure Formation
1.
2.
3.
4.
5.
6.
7. Orb[i]s | AADRL
Thesis | 2018-2020
130
Structure Formation This illustrates the transformative nature of the pyramidal system by virtue of connection and contraction of the legs. 32 units
1.
2.
3.
4.
5.
6.
7.
8.
Structure Formation
Temporary Scaffolds in Structure
Studies into the creation of a temporary structural elements which could aid in the formation of a structure, was also studied. 805 units
1.
2.
3.
4.
5.
6.
7.
8.
Orb[i]s | AADRL
Thesis | 2018-2020
132
Structure Formation Gradient Adaptation Following Gradient
1. Target destination on a gradient
2. Units follow the gradient
3. Units adapt and connect
4.Formation along gradient
Not Following Gradient
1.Units come together
2. Units climb over one another
3. Units maintain level, not following gradient
4. The collective of the units maintain one level.
Structure Formation
Following Gradient
1.Units add on to an existing aggregation
2. Level varies while climbing occurs
3. Levels vary in the process of climbing
4.Original level restored
Gradient Adaptation
5.
Orb[i]s | AADRL
Thesis | 2018-2020
134
Goal Oriented Behaviour Different goals of aggregating linearly, forming seating, vertical aggregation and forming a stepped structure are explored here.
1.
2.
3.
4.
5.
6.
7.
8.
Structure Formation
Linear Aggregation
Stepped Structure
Vertical Aggregation
Seating Formation Orb[i]s | AADRL
Thesis | 2018-2020
136
Structural Typologies
Number of Units: 40 State : -1
Number of Units: 40 State : 1
Number of Units: 40
Structure Formation
Number of Units: 55
Number of Units: 55
Number of Units: 55 Orb[i]s | AADRL
Thesis | 2018-2020
138
Structural Typologies
Number of Units: 90
Number of Units: 60
Number of Units: 54
Structure Formation
Number of Units: 110
Number of Units: 160
Number of Units: 275 Orb[i]s | AADRL
Thesis | 2018-2020
140
Behaviour
Goal-Oriented Behaviour Singular Target
Active Unit
Target
1.
2.
3.
4.
5.
6. The active unit emitting the signal moves towards the target location. The other units follow the active unit to ultimately reach the target.
Behaviour
5 Different Targets
Targets
1.
Units
5.
2.
6.
3.
7.
4. 8. The units aggregate towards the corresponding target points of the same color to form the desired structure. Orb[i]s | AADRL
Thesis | 2018-2020
150
Goal-Oriented Behaviour Multiple Obstacle Avoidance
Target
1.
Obstacles Units
5.
2.
6.
3.
7.
4. 8. The units learn to avoid and navigate through obstacles to reach the target location.
Behaviour
5 Different Targets & Obstacle Avoidance
Target
Obstacles
Units
1.
6.
2.
7.
3.
8.
4.
9.
5. 10. The units aggregate towards the corresponding target points of the same color to form the desired structure while avoiding obstacles. Orb[i]s | AADRL
Thesis | 2018-2020
152
Goal-Oriented Behaviour Overcoming uneven terrain and structure formation
Target
1.
2.
Units
Obstacle
4.
5.
3. 6. The units learn to manouver over uneven terrain to reach the target location and build the desired structure.
Behaviour
Orb[i]s | AADRL
Thesis | 2018-2020
Sensing & Communication Signal Detection | 1 Signal
1. Signal Initiated in the Active unit
Proximity Detection
2. Other units detect the signal
Direction Navigation
3. Units seek and aggregate towards the signal
Behaviour
Obstacle Avoidance
Target
1.
Unit
Target activated
Obstacle & Neighbor Avoidance
Neighbor Avoidance
1.
Behavior
Target
Unit
Unit Obstacle
2.
Unit Avoids Obstacle
2.
Unit start moving towards Target
3.
Unit moves towards Target
3.
More units moves towards Target
4.
Unit moves towards Target
4.
Units move towards Target
5.
Unit reaches Target
5.
All units reach Target
Orb[i]s | AADRL
Thesis | 2018-2020
156
Sensing & Communication Signal Detection | 3 Signals
Signal 1
Signal 2
Signal 1 & 2 activated
Signal 3
Signal 3 activated and units move towards Signal 1 & 2
Behaviour
Units start to seek and aggregate towards signals 1 & 2 & 3
Units aggregate towards the signals Orb[i]s | AADRL
Thesis | 2018-2020
158
Communication Transfer of information
Within a self-structuring and self-organizing system, the transfer of information is key. A constant loop of to and fro information is essential in order for the system to respond promptly to the changes within and outside the system.
Stimulus
Message
Receptor
Response
Message
Control Apparatus
Stimulus
Feedback Loop In our system, information loops are set up in at two levels: - Local Level At this level responses would be stimulated within the system at a more regional level, evoking responses from individual units These stimuli are generally territorial and are evoked based on their neighbouring units or immediate context. - Global Level At the global level stimulus is provided to the system at large. Two kinds of stimuli are provided to the system one based on context and environmental conditions and one through a cloud based server. The former is still a local feedback loop within the system, its the collective response of the unit that makes the response seem like a global response. Whereas, the latter information loop will be used to aid the location of the units in helping tackle dinginess, inactivity and induce a sense of safety. Movement and illumination would be provided in areas that are marked unsafe, promoting safer access and a psychological comfort to pedestrians accessing these spaces.
Behaviour
Locally
Environmental changes Movement patterns
Passive
Neighbouring units
Stimulus
Active
Feedback Loop
Globally
Orb[i]s | AADRL
Thesis | 2018-2020
160
Sensing & Communication Multiple Targets & Obstacle Avoidance Static Obstacles
Dynamic Obstacles
Inactive Units Active Units
Units and Obstacles in a given field
3 Targets activated Artificial Intelligence is used to stimulate the inactive units at random trajectories to reach the closest target, resulting in a different outcome each time. AI also enables a real-time avoidance of Static and Dynamic obstacles.
Behaviour
Units learn to avoid Static and Dynamic Obstacles and try to reach Targets
Units reach their Targets
Orb[i]s | AADRL
Thesis | 2018-2020
162
Sensing & Communication Range Finding & Obstacle Avoidance Static Obstacles
Dynamic Obstacles
Inactive Units Active Units
Target, Units and Obstacles in a given field
Units get activated when in range of Target The target unit is programmed with the capacity of range finding and a fixed angle for its radar detection.
Behaviour
In Range < 20 Out of Range > 20
Other Units get activated (grey -> orange) when in range of Target (distance < 20)
More Units within the range of the Target
Orb[i]s | AADRL
Thesis | 2018-2020
164
Sensing & Communication Signal Transmission & Obstacle Avoidance Static Obstacles
Dynamic Obstacles
Inactive Units Active Units
1. Passive and active agents, with obstacles in a given environment.
2. Passive agents get activated when in range of active agents. Agents learn to avoid obstacles with their angle detection capability.
Behaviour
In Range < 20 Out of Range > 20
Angle of Detection : 35°
3. Upon activation through range finding, agents started to display a clustering behavior with fellow active agents. Total time taken : 7 minutes 35 seconds
4. The total time taken for the activation of all agents in a given environment varies from agent speed and range of detection.
Orb[i]s | AADRL
Thesis | 2018-2020
166
Physical Sensing Prototype Inner Rolling Mechanism Exploded Axonometric
Outer Shield
Motor Shield
[2 x L293D Motor Driver]
Arduino Uno R3
Soft Silicon Wheel [38 x 11mm] 6V Batteries Micro DC Motor
[Gear Box | Speed Reducing] Gear Ratio Free-Run Speed Free-Run Current Motor Type Size Weight Shaft Diameter
50:1 30 rpm 120 mA 1.6 A Stall @6V (HP 6V) 35 x 12 x 15 mm 9.5 g 3 mm
Goal-Oriented Behaviour Obstacle Avoidance
Blue: No obstacle detected
Red: Obstacle detected
Behaviour
Light Sensing
Red: Unit Inactive
Blue: Unit Activated
Orb[i]s | AADRL
Thesis | 2018-2020
170
Goal-Oriented Behaviour Singe Target
Active Unit
Target
1. Target detected
Active Unit
Target
2. Move towards target
Active Unit Target
3. Target achieved
Behaviour
Colour Based Behaviour
Target
Active Unit
[In Range]
Target
Target
Target Detected
Target Achieved
Orb[i]s | AADRL
Thesis | 2018-2020
172
Collective Intelligence
Machine Learned Systems Initial Research What is Machine Learning? Machine Learning is an application of artificial intelligence that provides a system with the ability of learning from data autonomously, rather than in an engineered way. This works by feeding the system with information and observations that can be used to find patterns and make predictions about future outcomes. More broadly, the system has to learn the desired input-output mapping. This way, the system will be able to choose the best action to perform next in order to optimize its outcome. There are different ways of doing this depending on what kind of observations we have available to feed our system. The one used in this context is Reinforcement Learning. Reinforcement Learning is a learning method based on not telling the system what to do, but only what is right and what is wrong. This means that we let the system perform random actions and we provide a reward or a punishment when we think that these actions are respectively right or wrong. Eventually, the system will understand that it has to avoid making mistakes and earn maximum rewards. We most commonly opt for Reinforcement Learning to train a Neural Network, which is a computer system modeled on the nervous system.
Neural Networks: a Computerâ&#x20AC;&#x2122;s Nervous System
Collective Intelligence
In Unity, our â&#x20AC;&#x153;agentsâ&#x20AC;? (the entities who perform the learning) use the Reinforcement Learning model. The agents perform actions in a given environment. The actions cause a change in the environment, which in turn is feed back to the agent, together with some reward or punishment. The action happens in what we call the Learning Environment. State & Rewards
State Transition
Environment
Agent
Action
Machine Learning is changing the way we expect to get intelligent behaviour out of autonomous agents. Whereas in the past the behaviour was coded by hand, it is increasingly taught to the agent (either a robot or virtual avatar) through interaction in a training environment. This method is used to learn behaviour for everything from industrial robots, drones, and autonomous vehicles, to game characters and opponents. The quality of this training environment is critical to the kinds of behaviours that can be learned, and there are often trade-offs of one kind or another that need to be made. Machine Learning becomes integral in our system to be able to achieve efficient autonomy. The unique and non-predefined behaviour in the system would arise from this kind of decision making. The agents in a given environment will learn from its experience and ensure to perform a more efficient action each time, making the system smart, behaviour driven and autonomous.
Orb[i]s | AADRL
Thesis | 2018-2020
178
Q-Learning
Within the scope of Reinforcement Learning, Q-Learning provides us a function which helps estimate long term values of taking certain actions in certain circumstances. The goal with Reinforcement Learning is to train an agent which can learn to act in ways that maximizes future rewards within a given environment. Q-Learning finds the optimal path to the goal when the path is unknown/ cannot be predicted. The primary difference in Q-Learning and Reinforcement Learning is that of the ‘contextual bandit’ working in Q-learning i.e., there is a lack of sparse rewards and state transition when compared to the latter. The agent doesn’t receive a reward for each action taken, rather waits till an optimal number of actions have been taken. This estimate value improves with time, optimizing the actions taken.
Testing the Agents in a Grid World of Q-Learning An environment with: • • • •
Agents : randomly placed Goals : randomly placed - position we wish for the agents to learn of Obstacles : randomly placed - which we want our agents to avoid States : in the environment that will correspond to the grid.
At the inception of the training the Q-value estimate is poor. The Q-value corresponds to the true Q-function of the environment, therefore, actions taken tend to be increasingly accurate. The point based system of learning aids the reward-punishment learning system. Points: • +1 : for moving towards the goal • - 1 : for moving towards an obstacle • - 0.05 : for each move made (encouraging quicker movement).
Collective Intelligence
Orb[i]s | AADRL
Thesis | 2018-2020
180
Collective Intelligence Behaviour Designer
In order to study Collective Intelligence in our system with efficient Machine Learning, behavioural studies became integral. Behaviour of agents lay the basis of Collective Artificial Intelligence. Thus, by using Behaviour trees in our simulations, we were able to achieve unpredicted AI. Each agent within a system behaves uniquely, having different goals and different ways of achieving them. But what happens when 100’s of these agents come together? Each agent’s behaviour contributes to a collective whole behaviour. In order to understand this collective behaviour better, we did several tests with a varied set of behaviours. One of our initial tests was the ‘Seeker-Attractor’ Behaviour, where one agent would have the property to seek and another to attract the seeker. By merely varying the proportions of Seeker to Attractors within a given environment, we observed a large variation in clustering patterns and formations. Rules : Attractor
Seeker
• Attractors freely navigate a given space, taking random paths each time - much like analyzing the environment. • Attractors repel each other when within a distance ‘X’. • Seekers remain stationery unless they come within range ‘Y’ of an Attractor. • Seekers keep count of the number of collisions with an attractor, using which an agent determines which attractor it can establish a stronger connection with.
Collective Intelligence
1 Attractor 5 Seekers
2 Attractors 10 Seekers
3 Attractors 30 Seekers
4 Attractors 40 Seekers
5 Attractors 50 Seekers
6 Attractors 60 Seekers
Common Cluster formations observed
Interaction between attractors and seekers
Orb[i]s | AADRL
Thesis | 2018-2020
182
Collective Intelligence Detecting Neighbouring Agents
Establishing a Relationship In order to develop collective behaviour, agents must have the ability to sense and detect fellow agents in a given context. An apt angle of detection as well as range of detection is key for identification of neighbours followed by establishment of suitable connections. By providing agents with a 360 degree sensing ability, the connections with other agents are then not limited to a particular direction. To understand this relationship better, we treated our agents as particles. Parameters like Velocity, Gravity and Particle Distances were predetermined to get optimal results. By varying the number of agents in the same given space, a large number of patterns in raycasts were observed, much like a large interconnected network. In this case the Optimal values for interaction were fixed at : â&#x20AC;˘ Range : 0 - 85 units â&#x20AC;˘ Agent Velocity : Velocity_Y direction + 0.8
5 Agents
10 Agents
20 Agents
50 Agents
100 Agents
200 Agents
500 Agents
1000 Agents
Collective Intelligence
Flocking / Swarming Behaviour Flocking Behaviour was introduced in the system by applying and varying the following forces. A different and unpredicted pattern was observed with variation in these forces as well as with number of agents present.
Alignment
Cohesion
Alignment enables the movement of agents in a common direction. Each heading will set towards an average heading direction.
Cohesion compels each agent in a flock to stay grouped with its neighbours. Agent finds a mid point between all its neighbours and then navigates towards it.
Separation
Separation helps prevent collisions and overlaps of agents. If other agents get too close in radius, the agent will navigate away from these neighbours.
5 Agents
10 Agents
20 Agents
50 Agents
100 Agents
200 Agents
500 Agents
1000 Agents
Alignment : 0.8 Cohesion : 0.15 Separation: 0.25
Orb[i]s | AADRL
Thesis | 2018-2020
184
Collective Intelligence Units aggregating in a single layer
Collective Intelligence
Units aggregating to grow vertically
Orb[i]s | AADRL
Thesis | 2018-2020
186
Stigmergy In higher population studies, these units depict intelligent behaviors in flocking strategies. They begin to display stigmergic properties wherein the agents sense and follow the other units closer to the target. Constrained by the axial mobility properties of the unit.
1750 Units | 1 Flock
Collective Intelligence
Target
Orb[i]s | AADRL
Thesis | 2018-2020
188
Stigmergy These flocks also collectively move as per the displacement, while being constrained by the axial mobility properties of the unit.
3000 Units | 2 Flocks
Collective Intelligence
Target
Orb[i]s | AADRL
Thesis | 2018-2020
190
Crowd Segregation Depending on the number of units present in a given location, they divide the space differently resulting in spaces of different user occupancies.
50 Units
200 Units 10-40 Person Occupancy
40-100 Person Occupancy
Collective Intelligence
2000 Units
2000 Units Movement Patterns
Orb[i]s | AADRL
Thesis | 2018-2020
192
Path Formation Upon contact with a human agent, the units self-organize to move away and clear the path taken by the human.
2000 Units
Collective Intelligence
Orb[i]s | AADRL
Thesis | 2018-2020
194
Prototyping
Prototype Detail Prototype Components
Exoskeleton
Extendable Arm Mechanism
Connection Point
Inner Rolling Mechanism
Prototyping
125
Outer Surface Geometry Initial Exoskeleton Prototypes Pros: Symmetrical Light weight
Cons: Less Curvature Inability to roll
Pros: Increased Curvature Light weight
Cons: Asymmetrical Inability to roll
Pros: Increased Curvature
Cons: Asymmetrical leading to undistributed weight Inability to roll
Prototyping
Pros: Increased Curvature
Cons: Asymmetrical Heavy Difficulty Rolling
Pros: Increased Curvature Light-weight Symmetrical
Cons: Difficulty Rolling smoothly
Pros: Increased Curvature Very Light-weight Symmetrical
Cons: Difficulty Rolling smoothly
Orb[i]s | AADRL
Thesis | 2018-2020
202
Prototype Exoskeleton Detail Mobility Test
Subdivisions Level: 1
Ease Of Rolling
Subdivisions Level: 2
Subdivisions Level: 3
Subdivisions Level: 4
Prototyping
Kerfing Pattern on Rubber Track
Rigid Frame Indent on geometry for a flushed track
Geodesic Triangular Geometry
Orb[i]s | AADRL
Thesis | 2018-2020
206
Prototype Exoskeleton Detail
Force Transfer Compression Ring
Addition of compression rings to stabilize nodes
Increasing Triangular Subdivisions
Increasing Triangular Subdivisions
Prototyping
Ax8
Ex8
Bx4
Fx8
Cx8
Gx4
Dx8
Hx8
Orb[i]s | AADRL
Thesis | 2018-2020
210
Materiality Recycling - Upcycling Itâ&#x20AC;&#x2122;s a known fact that plastics are hard on the environment. While several organizations and establishments have recycling programs, the amount of plastic produced each year overshadows the amount that gets recycled. Significant amount of energy goes into the collection and processes of recycling, which nullifies its purpose. Moving past the entire process of collection and commuting, 3D printing has disrupted these barriers. With its ability to decentralize manufacturing, parts can be emailed and printed on location. Some companies, such as b-pet, ProCycler and Filabot are making desktop solutions to grind and extrude your own polymer practically anywhere. Recently, researchers at Michigan Technological University in Houghton Mich., released all the open-source files online to build a machine that can recycle polymers and turn them into filament.
Plastic Waste
ProtoCycler : Plastic
3D Printing : Prototype
One method of distributed plastic recycling is to upcycle plastic waste into 3D printing filament with a recyclebot, which is an open-source waste plastic extruder. Previous research on the life cycle analysis (LCA) or the recyclebot process using post-consumer plastics instead of raw materials, showed a 90% decrease in the embodied energy of the filament from the mining, processing of natural resources, and synthesizing compared to traditional manufacturing.
Prototyping
In addition, the recyclebot provides the potential for consumers to recycle plastic in their own homes to save money by offsetting purchased filament. Recyclebots are also useful for laboratory and industry prototyping research, as failed prototypes are recycled into filament for future work. There have been many versions of recyclebot developed by both companies (e.g., Filastruder) as well as individuals (e.g., Lyman), including open-source versions from the Plastic Bank, Precious Plastic, and Perpetual Plastic. There are also several commercial versions of the recyclebot, including the Filastruder, Filafab, Noztek, Filabot, EWE, Extrusionbot, Filamaker (which also has a shredder), Strooder, and Felfil (OS), that can work with waste plastic.
Orb[i]s Out in the real world
Wear and Tear
Orb[i]s Out in the real world
Orb[i]s | AADRL
Thesis | 2018-2020
212
Recharging Potential Our system is incorporated with the ability to recharge itself with wireless inductive charging using the existing infrastructure in London for this purpose. For this purpose we studied the availability of these charging stations around London and found them to be in a convenient 1 mile radius.
Conserving Energy
The kinetic energy generated from braking is converted to thermal energy and reused to power the unit.
Prototyping
Inductive Recharging Mechanism
Charging up in Different Seasons
Battery charge lasts longer in summer months due to warmer ambient temperature
Summer
Winter
Orb[i]s | AADRL
Thesis | 2018-2020
214
2
Prototype Detail Inner Rolling Mechanism Exploded Axonometric
Prototyping
Wheels DC Motors Counterweight
Orb[i]s | AADRL
Thesis | 2018-2020
218
Prototype Detail Extendable Arm Mechanism Exploded Axonometric
Electric Actuator
Actuator Holder [3D Printed]
Inner Arm Tube
[6 cm Diameter | Acrylic]
Inner Arm Cap [3D Printed]
Prototyping
Electromagnet Spring
[3W | 0.13A]
Electromagnet Holder [3D Printed]
Flexible Arm Tube [4 cm Diameter]
Orb[i]s | AADRL
Thesis | 2018-2020
222
Prototype Detail Extendable Arm Mechanism Exploded Axonometric
Inner Arm Tube
[6 cm Diameter | Acrylic]
Motor Holder [3D Printed]
Micro DC Motor
[Gear Box | 12cm Shaft] Gear Ratio Free-Run Speed Free-Run Current Motor Type Size Weight Shaft Diameter Output Shaft Length
Inner Arm Cap [3D Printed]
50:1 400 rpm 170 mA 1.6 A Stall @12V (HP 12V) 144 x 12 x 10 mm 20 g 4 mm 120 mm
Prototyping
Ring Holder [3D Printed]
Outer Arm Tube
[4cm Diameter | Acrylic]
Electromagnet [3W | 0.13A]
Load Capacity Current Magnet Type Size Weight Mounting Hole
2.5 Kg 130 mA EM200815M3-5(HP 12v) 15 x 20 mm 150 g 3 mm
Electromagnet Holder [3D Printed]
Orb[i]s | AADRL
Thesis | 2018-2020
224
Detail Drawing Exploded Axonometric | Design 2
2
Unit to Uni
it connection
Flexible Extending Tube
Illumination Strategies
System Intelligence Detecting and Recording Motion Patterns
By using Passive Infrared (PIR) sensors, motion of users can be recorded within the system. Furthermore, by recording the start and end time of the motion the duration of motion can be estimated. With this information, the system would have the ability to study and analyze user patterns and bring about meaningful transformations in the system, making the spaces further conducive. Code used to record data using PIR Sensor : void loop() { val = digitalRead(inputPin); if (val == HIGH) { // checking if the input is HIGH digitalWrite(ledPin, HIGH); // turns LED ON if (pirState == LOW) { Serial.println(“Motion detected!”); pirState = HIGH; } } else { digitalWrite(ledPin, LOW); // turns the LED OFF if (pirState == HIGH){
}
}
Serial.println(“Motion ended!”); pirState = LOW;
Motion detected and recorded
Motion detected using PIR sensor
Illumination Strategies
With the use of Playmaker’s components, the PIR sensors capacity to record motion could be recreated and studied further in our simulations. Using the ‘ Counter’ feature on each unit of the system, the change in data could be recorded to bring about required reconfigurations.
PlayMaker Script for ‘Counter’
Counter Initiation with motion of passers-by
Counter Incremented with further motion of passers-by
Orb[i]s | AADRL
Thesis | 2018-2020
244
System Intelligence Detecting and Recording Motion Patterns
Movement Curves
1 Person
3 People
7 People
Illumination Strategies
By adopting the Counter feature in simulations, movement patterns could be recorded. These Movement Curves could be used to trigger transformations within a formed space. Real-Time Movement Pattern Analysis
Coinciding Patterns
Coinciding Patterns Orb[i]s | AADRL
Thesis | 2018-2020
246
09.01.2020
Bibliography
Website References • "Brume". 2012. Studio Joanie Lemercier. https://joanielemercier. com/brume/. • "Light Pollination — High Lumen". 2020. High Lumen. https://www. highlumen.me/blog/2018/01/13/light-pollination. • "Mapping London". 2020. Mappinglondon.Co.Uk. https://mappinglondon. co.uk/. • Westland, Stephen. 2017. "Does Colour Really Affect Our Mind And Body? A Professor Of Colour Science Explains". Blog. The Conversation. https://theconversation.com/does-colour-really-affect-ourmind-and-body-a-professor-of-colour-science-explains-84382. • "The Psychological Impact Of Light And Color - TCP Lighting". 2015. TCP Lighting. https://www.tcpi.com/psychological-impact-light-color/. • Herzog, Katie. 2013. "Why Shade Balls Aren’T Such A Great Idea After All". Grist. https://grist.org/article/why-shade-balls-arentsuch-a-great-idea-after-all/.
272
Book References • Alpaydin, Ethem. 2016. Machine Learning. Cambridge, MA : MIT Press, [2016]. • Brown, Henry T. n.d. 507 Mechanical Movements - Mechanisms And Devices. New York: Dover Publications, 2005. • Downing, Keith L. 2015. Intelligence Emerging. 12th ed. Cambridge, Massachusetts: The MIT Press, [2015]. • Engelbrecht, Andries P. 2007. Fundamentals Of Computational Swarm Intelligence. Chichester: John Wiley. • Gehl, Jan, and Jo Koch. 2011. Life Between Buildings. London: Island Press • Hall, Edward T. 1990. The Hidden Dimension. New York: Anchor Books. • Hensel, Michael. 2004. Emergence. Chichester: Wiley-Academy. • Holland, John H. 2010. Emergence. Oxford: Oxford Univ. Press. • Ijeh, I. (2018). Spherical objects!. [online] Building. Available at: https://www.building.co.uk/buildings/spherical-objects/5092601. article [Accessed 24 Dec. 2019]. • Johnson, Steven. 2001. Emergence. London: Allen Lane The Penguin Press. • Kennedy, James, Russell C Eberhart, and Yuhui Shi. 2009. Swarm Intelligence. San Francisco: Kaufmann. • Lanham, Michael. n.d. Learn Unity ML-Agents. Birmingham: Packt Publishing, 2018. • Malone, Thomas W. 2015. Handbook Of Collective Intelligence. Cambridge, MA: The MIT Press. • Oxer, Jonathan. 2010. Practical Arduino. [Berkeley, CA]: Apress.
Orb[i]s | AADRL
Thesis | 2018-2020