SlideShare a Scribd company logo
1 of 16
Download to read offline
4.7-9 EuroGP2021@Seville, Spain
Regenerating Soft Robots
through Neural Cellular Automata
Kazuya Horibe (@khoribe3)1,2,3 , Kathryn Walker1 , Sebastian Risi1
1 IT University of Copenhagen
2 Osaka University
3 Cross Labs, Cross Compass Ltd.
1
2
Background
Morphological regeneration is an important feature that highlights
the environmental adaptive capacity of biological system.
[Mitoh+2021, Current Biology]
Extreme autotomy and whole-body regeneration in photosynthetic sea slugs.
3
Lack of this regenerative capacity significantly limits the resilience of
machines and the environments they can operate in.
Can we give regenerative capacity to soft robots?
Background
[Shah+2020, Advanced Materials]
Regeneration
[Mitoh+2021, Current Biology]
Regeneration can make a new dimension of freedom for shape.
Machines Biological System
[Kriegman+2020, PNAS]
Soft robots using biological tissues
Hybrid System
XenoBod
4
Method: Soft robots simulator
We propose a model for soft robots that regenerate through
a neural cellular automata.
evosoro: soft robot simulator
• Python soft robot simulation library based on the Voxelyze physics engine. 
• High-level interface for the dynamic simulation and automated design of soft multimaterial robots.
[Hiller+2014, Soft Robotics]
[Cheney+2013, GECCO]
Type of voxels
5
[Mordvintsev+2020, Distill]
Applying neural cellular automata (NCA) to growth and
regeneration rule of soft robots.
Method: Neural Cellular Automata
NCA can learn growth and regenerative rule. Applying NCA to 2D soft robots
[Nichele+2017, IEEE Transactions on
Cognitive and Developmental Systems]
Feed forward (Linear unit)

Recurrent (LSTM unit)
Network structure
Hidden layer: 64 hidden state
[Hochreiter1997, Neural computation]
Input layer: Number of neighborhood
Output layer: Number of voxel states
6
Method: Training for locomotion
Training for locomotion of soft robots through genetic algorithm
Get phenotype
(Morphology of soft robots)
Locomotion Evaluation
In a constant time
Fitness value
(Travel distance of Center of Mass)
[Such+2017, arXiv preprint]
Evolve weight of genotype (NCA)
Grow soft robots through NCA from single voxel (10 steps)
7
Research strategy
Growth rule for 2D soft robots
Growth rule for 3D soft robots
Regenerative rule for 3D soft robots
We investigate that neural cellular automata can learn growth and
regenerative rule for soft robots.
8
Result 1
2D soft robots get various morphology through evolution.
Training curve
10 independent runs each condition
We confirmed NCA works well for growing rule of soft robots.
Feed forward
Feed forward Recurrent
Recurrent
300 population and 500 generation
Common grown soft robots
S-type
Biped L-type
Zigzag
9
Training curve
24 independent runs each condition
We tried a more complex task: 3D soft robots locomotion
Result 1
Feed forward
Feed forward
Feed forward
Feed forward
Feed forward
Recurrent
Recurrent Recurrent
Recurrent
Feed forward
2D Group 3D Group
100 population and 300 generation
Common grown soft robots
Roller
Jumper
Pull-Push
Slider
Jitter
Crawler
Slider
Pull-Push
L-Walker
Jumper
10
Partial Summary
Partial Summary for growth capacity of soft robots
We gave growth capacity to 2D and 3D soft robots through neural
cellular automata.
An advantage of the presence of cellular memory (LSTM unit) was
depend on tasks.
A morphology (e.g. Tripod) that did not appear in previous studies using
CPPN [Cheney+2013, GECCO] was acquired.
11
We investigate the ability of the soft robots to regenerate their body
parts to recover from morphological damage.
NCAs that are trained for locomotion failed regeneration.
Result 2
Regrowing 10 steps from damaged morphology
12
Training for regeneration of soft robots through genetic algorithm
Get phenotype
(Morphology of soft robots)
Similarity Evaluation
between the original morphology
Fitness value
(Similarity)
[Such+2017, arXiv preprint]
Evolve weight of genotype (NCA)
Grow soft robots through NCA from damaged morphology (10 steps)
Method: Regeneration task
Regrown Original
Compare position (x, y, z) and type of voxel (Muscle, Tissue, Bone, Empty) for 9*9*9=729 voxels.
13
Soft robots are almost regrown through NCA.
Result 2
Training curve
1 independent run
1000 population and 1000 generation
100% (729/729) is completely consistent with the original morphology.
14
Soft robots are partially recover their locomotion.
Result 2
Tripod
Travel distance
40.4 1.63 (3.6%) 20.3 (45%)
15
Summary
Summary
We gave growth capacity to soft robots through neural cellular automata.
An advantage of the presence of cellular memory (LSTM unit) is depend on tasks in
our experiments.
A morphology (e.g. Tripod) that did not appear in previous studies using CPPN
[Cheney+2013, GECCO] was acquired.
We gave regenerative capacity to soft robots through neural cellular automata.
This model requires two different neural cellular automata for growth and regeneration.
16
Discussion
[Kriegman+2020, PNAS]
Biological organisms can memorize a global morphology through bioelectric signaling network.
[Levin+2017, Annual Review of Biomedical Engineering]
Bioelectric signaling network can modify using extrinsic electrical stimuli.
Hybrid soft robots may realize the regenerative ability of machines.
XenoBod

More Related Content

Similar to 210408EuroGP

MaartenDekker_Mechanical_design_of_a_humanoid_robots_lower_body
MaartenDekker_Mechanical_design_of_a_humanoid_robots_lower_bodyMaartenDekker_Mechanical_design_of_a_humanoid_robots_lower_body
MaartenDekker_Mechanical_design_of_a_humanoid_robots_lower_bodyMaarten Dekker
 
5 multi robot path planning algorithms
5 multi robot path planning algorithms5 multi robot path planning algorithms
5 multi robot path planning algorithmsprjpublications
 
Analytical Development of the Forward and Inverse Kinematics of A Robotic Leg...
Analytical Development of the Forward and Inverse Kinematics of A Robotic Leg...Analytical Development of the Forward and Inverse Kinematics of A Robotic Leg...
Analytical Development of the Forward and Inverse Kinematics of A Robotic Leg...IJERA Editor
 
2016 Summer Research Report
2016 Summer Research Report2016 Summer Research Report
2016 Summer Research ReportJames Boggs
 
Swarm intelligence pso and aco
Swarm intelligence pso and acoSwarm intelligence pso and aco
Swarm intelligence pso and acosatish561
 
Comparison Between PSO and HPSO In Image Steganography
Comparison Between PSO and HPSO In Image SteganographyComparison Between PSO and HPSO In Image Steganography
Comparison Between PSO and HPSO In Image SteganographyIJCSIS Research Publications
 
BIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOT
BIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOTBIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOT
BIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOTJaresJournal
 
BIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOT
BIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOTBIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOT
BIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOTJaresJournal
 
adaptive comprehensive learning bacterial foraging optimization and its appli...
adaptive comprehensive learning bacterial foraging optimization and its appli...adaptive comprehensive learning bacterial foraging optimization and its appli...
adaptive comprehensive learning bacterial foraging optimization and its appli...Amir Shokri
 
Learning of robot navigation tasks by
Learning of robot navigation tasks byLearning of robot navigation tasks by
Learning of robot navigation tasks bycsandit
 
LEARNING OF ROBOT NAVIGATION TASKS BY PROBABILISTIC NEURAL NETWORK
LEARNING OF ROBOT NAVIGATION TASKS BY PROBABILISTIC NEURAL NETWORKLEARNING OF ROBOT NAVIGATION TASKS BY PROBABILISTIC NEURAL NETWORK
LEARNING OF ROBOT NAVIGATION TASKS BY PROBABILISTIC NEURAL NETWORKcsandit
 
Learning Social Affordances and Using Them for Planning
Learning Social Affordances and Using Them for PlanningLearning Social Affordances and Using Them for Planning
Learning Social Affordances and Using Them for PlanningKadir Uyanik
 
01 introduction
01 introduction01 introduction
01 introductionHau Nguyen
 
Spike timing dependent plasticity to make robot navigation more intelligent. ...
Spike timing dependent plasticity to make robot navigation more intelligent. ...Spike timing dependent plasticity to make robot navigation more intelligent. ...
Spike timing dependent plasticity to make robot navigation more intelligent. ...Lietuvos kompiuterininkų sąjunga
 
Particle swarm optimization
Particle swarm optimizationParticle swarm optimization
Particle swarm optimizationMahesh Tibrewal
 
EFFECTS OF THE DIFFERENT MIGRATION PERIODS ON PARALLEL MULTI-SWARM PSO
EFFECTS OF THE DIFFERENT MIGRATION PERIODS ON PARALLEL MULTI-SWARM PSOEFFECTS OF THE DIFFERENT MIGRATION PERIODS ON PARALLEL MULTI-SWARM PSO
EFFECTS OF THE DIFFERENT MIGRATION PERIODS ON PARALLEL MULTI-SWARM PSOcscpconf
 

Similar to 210408EuroGP (20)

MaartenDekker_Mechanical_design_of_a_humanoid_robots_lower_body
MaartenDekker_Mechanical_design_of_a_humanoid_robots_lower_bodyMaartenDekker_Mechanical_design_of_a_humanoid_robots_lower_body
MaartenDekker_Mechanical_design_of_a_humanoid_robots_lower_body
 
5 multi robot path planning algorithms
5 multi robot path planning algorithms5 multi robot path planning algorithms
5 multi robot path planning algorithms
 
Analytical Development of the Forward and Inverse Kinematics of A Robotic Leg...
Analytical Development of the Forward and Inverse Kinematics of A Robotic Leg...Analytical Development of the Forward and Inverse Kinematics of A Robotic Leg...
Analytical Development of the Forward and Inverse Kinematics of A Robotic Leg...
 
2016 Summer Research Report
2016 Summer Research Report2016 Summer Research Report
2016 Summer Research Report
 
Robotics report
Robotics reportRobotics report
Robotics report
 
Swarm intelligence pso and aco
Swarm intelligence pso and acoSwarm intelligence pso and aco
Swarm intelligence pso and aco
 
November 16, Learning
November 16, LearningNovember 16, Learning
November 16, Learning
 
Comparison Between PSO and HPSO In Image Steganography
Comparison Between PSO and HPSO In Image SteganographyComparison Between PSO and HPSO In Image Steganography
Comparison Between PSO and HPSO In Image Steganography
 
BIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOT
BIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOTBIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOT
BIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOT
 
BIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOT
BIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOTBIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOT
BIO-INSPIRATIONS AND PHYSICAL CONFIGURATIONS OF SWARM-BOT
 
adaptive comprehensive learning bacterial foraging optimization and its appli...
adaptive comprehensive learning bacterial foraging optimization and its appli...adaptive comprehensive learning bacterial foraging optimization and its appli...
adaptive comprehensive learning bacterial foraging optimization and its appli...
 
Learning of robot navigation tasks by
Learning of robot navigation tasks byLearning of robot navigation tasks by
Learning of robot navigation tasks by
 
LEARNING OF ROBOT NAVIGATION TASKS BY PROBABILISTIC NEURAL NETWORK
LEARNING OF ROBOT NAVIGATION TASKS BY PROBABILISTIC NEURAL NETWORKLEARNING OF ROBOT NAVIGATION TASKS BY PROBABILISTIC NEURAL NETWORK
LEARNING OF ROBOT NAVIGATION TASKS BY PROBABILISTIC NEURAL NETWORK
 
Learning Social Affordances and Using Them for Planning
Learning Social Affordances and Using Them for PlanningLearning Social Affordances and Using Them for Planning
Learning Social Affordances and Using Them for Planning
 
kfu_poster_a4
kfu_poster_a4kfu_poster_a4
kfu_poster_a4
 
dheeraj
dheerajdheeraj
dheeraj
 
01 introduction
01 introduction01 introduction
01 introduction
 
Spike timing dependent plasticity to make robot navigation more intelligent. ...
Spike timing dependent plasticity to make robot navigation more intelligent. ...Spike timing dependent plasticity to make robot navigation more intelligent. ...
Spike timing dependent plasticity to make robot navigation more intelligent. ...
 
Particle swarm optimization
Particle swarm optimizationParticle swarm optimization
Particle swarm optimization
 
EFFECTS OF THE DIFFERENT MIGRATION PERIODS ON PARALLEL MULTI-SWARM PSO
EFFECTS OF THE DIFFERENT MIGRATION PERIODS ON PARALLEL MULTI-SWARM PSOEFFECTS OF THE DIFFERENT MIGRATION PERIODS ON PARALLEL MULTI-SWARM PSO
EFFECTS OF THE DIFFERENT MIGRATION PERIODS ON PARALLEL MULTI-SWARM PSO
 

More from Kazuya Horibe

MBMWW2023_slideshare.pdf
MBMWW2023_slideshare.pdfMBMWW2023_slideshare.pdf
MBMWW2023_slideshare.pdfKazuya Horibe
 
220614MarguliesLab.pdf
220614MarguliesLab.pdf220614MarguliesLab.pdf
220614MarguliesLab.pdfKazuya Horibe
 
Curved surface geometry-induced topological change of an excitable planar wav...
Curved surface geometry-induced topological change of an excitable planar wav...Curved surface geometry-induced topological change of an excitable planar wav...
Curved surface geometry-induced topological change of an excitable planar wav...Kazuya Horibe
 
Morphogenesis in robot swarms
Morphogenesis in robot swarmsMorphogenesis in robot swarms
Morphogenesis in robot swarmsKazuya Horibe
 
Metabolic co-dependence gives rise to collective oscillations within biofilms.
Metabolic co-dependence gives rise to collective oscillations within biofilms.Metabolic co-dependence gives rise to collective oscillations within biofilms.
Metabolic co-dependence gives rise to collective oscillations within biofilms.Kazuya Horibe
 

More from Kazuya Horibe (6)

MBMWW2023_slideshare.pdf
MBMWW2023_slideshare.pdfMBMWW2023_slideshare.pdf
MBMWW2023_slideshare.pdf
 
220614MarguliesLab.pdf
220614MarguliesLab.pdf220614MarguliesLab.pdf
220614MarguliesLab.pdf
 
181226noneq
181226noneq181226noneq
181226noneq
 
Curved surface geometry-induced topological change of an excitable planar wav...
Curved surface geometry-induced topological change of an excitable planar wav...Curved surface geometry-induced topological change of an excitable planar wav...
Curved surface geometry-induced topological change of an excitable planar wav...
 
Morphogenesis in robot swarms
Morphogenesis in robot swarmsMorphogenesis in robot swarms
Morphogenesis in robot swarms
 
Metabolic co-dependence gives rise to collective oscillations within biofilms.
Metabolic co-dependence gives rise to collective oscillations within biofilms.Metabolic co-dependence gives rise to collective oscillations within biofilms.
Metabolic co-dependence gives rise to collective oscillations within biofilms.
 

Recently uploaded

Advantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your BusinessAdvantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your BusinessPixlogix Infotech
 
08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking MenDelhi Call girls
 
A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024Results
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationSafe Software
 
Histor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slideHistor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slidevu2urc
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonetsnaman860154
 
GenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationGenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationMichael W. Hawkins
 
Breaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountBreaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountPuma Security, LLC
 
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Igalia
 
Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024The Digital Insurer
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerThousandEyes
 
Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Enterprise Knowledge
 
Data Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonData Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonAnna Loughnan Colquhoun
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsMaria Levchenko
 
The Codex of Business Writing Software for Real-World Solutions 2.pptx
The Codex of Business Writing Software for Real-World Solutions 2.pptxThe Codex of Business Writing Software for Real-World Solutions 2.pptx
The Codex of Business Writing Software for Real-World Solutions 2.pptxMalak Abu Hammad
 
A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)Gabriella Davis
 
Scaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationScaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationRadu Cotescu
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityPrincipled Technologies
 
EIS-Webinar-Prompt-Knowledge-Eng-2024-04-08.pptx
EIS-Webinar-Prompt-Knowledge-Eng-2024-04-08.pptxEIS-Webinar-Prompt-Knowledge-Eng-2024-04-08.pptx
EIS-Webinar-Prompt-Knowledge-Eng-2024-04-08.pptxEarley Information Science
 
Automating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps ScriptAutomating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps Scriptwesley chun
 

Recently uploaded (20)

Advantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your BusinessAdvantages of Hiring UIUX Design Service Providers for Your Business
Advantages of Hiring UIUX Design Service Providers for Your Business
 
08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men08448380779 Call Girls In Civil Lines Women Seeking Men
08448380779 Call Girls In Civil Lines Women Seeking Men
 
A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024A Call to Action for Generative AI in 2024
A Call to Action for Generative AI in 2024
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
 
Histor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slideHistor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slide
 
How to convert PDF to text with Nanonets
How to convert PDF to text with NanonetsHow to convert PDF to text with Nanonets
How to convert PDF to text with Nanonets
 
GenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day PresentationGenCyber Cyber Security Day Presentation
GenCyber Cyber Security Day Presentation
 
Breaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountBreaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path Mount
 
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
 
Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024Finology Group – Insurtech Innovation Award 2024
Finology Group – Insurtech Innovation Award 2024
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected Worker
 
Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...
 
Data Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt RobisonData Cloud, More than a CDP by Matt Robison
Data Cloud, More than a CDP by Matt Robison
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed texts
 
The Codex of Business Writing Software for Real-World Solutions 2.pptx
The Codex of Business Writing Software for Real-World Solutions 2.pptxThe Codex of Business Writing Software for Real-World Solutions 2.pptx
The Codex of Business Writing Software for Real-World Solutions 2.pptx
 
A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)
 
Scaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationScaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organization
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
EIS-Webinar-Prompt-Knowledge-Eng-2024-04-08.pptx
EIS-Webinar-Prompt-Knowledge-Eng-2024-04-08.pptxEIS-Webinar-Prompt-Knowledge-Eng-2024-04-08.pptx
EIS-Webinar-Prompt-Knowledge-Eng-2024-04-08.pptx
 
Automating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps ScriptAutomating Google Workspace (GWS) & more with Apps Script
Automating Google Workspace (GWS) & more with Apps Script
 

210408EuroGP

  • 1. 4.7-9 EuroGP2021@Seville, Spain Regenerating Soft Robots through Neural Cellular Automata Kazuya Horibe (@khoribe3)1,2,3 , Kathryn Walker1 , Sebastian Risi1 1 IT University of Copenhagen 2 Osaka University 3 Cross Labs, Cross Compass Ltd. 1
  • 2. 2 Background Morphological regeneration is an important feature that highlights the environmental adaptive capacity of biological system. [Mitoh+2021, Current Biology] Extreme autotomy and whole-body regeneration in photosynthetic sea slugs.
  • 3. 3 Lack of this regenerative capacity significantly limits the resilience of machines and the environments they can operate in. Can we give regenerative capacity to soft robots? Background [Shah+2020, Advanced Materials] Regeneration [Mitoh+2021, Current Biology] Regeneration can make a new dimension of freedom for shape. Machines Biological System [Kriegman+2020, PNAS] Soft robots using biological tissues Hybrid System XenoBod
  • 4. 4 Method: Soft robots simulator We propose a model for soft robots that regenerate through a neural cellular automata. evosoro: soft robot simulator • Python soft robot simulation library based on the Voxelyze physics engine.  • High-level interface for the dynamic simulation and automated design of soft multimaterial robots. [Hiller+2014, Soft Robotics] [Cheney+2013, GECCO] Type of voxels
  • 5. 5 [Mordvintsev+2020, Distill] Applying neural cellular automata (NCA) to growth and regeneration rule of soft robots. Method: Neural Cellular Automata NCA can learn growth and regenerative rule. Applying NCA to 2D soft robots [Nichele+2017, IEEE Transactions on Cognitive and Developmental Systems] Feed forward (Linear unit) Recurrent (LSTM unit) Network structure Hidden layer: 64 hidden state [Hochreiter1997, Neural computation] Input layer: Number of neighborhood Output layer: Number of voxel states
  • 6. 6 Method: Training for locomotion Training for locomotion of soft robots through genetic algorithm Get phenotype (Morphology of soft robots) Locomotion Evaluation In a constant time Fitness value (Travel distance of Center of Mass) [Such+2017, arXiv preprint] Evolve weight of genotype (NCA) Grow soft robots through NCA from single voxel (10 steps)
  • 7. 7 Research strategy Growth rule for 2D soft robots Growth rule for 3D soft robots Regenerative rule for 3D soft robots We investigate that neural cellular automata can learn growth and regenerative rule for soft robots.
  • 8. 8 Result 1 2D soft robots get various morphology through evolution. Training curve 10 independent runs each condition We confirmed NCA works well for growing rule of soft robots. Feed forward Feed forward Recurrent Recurrent 300 population and 500 generation Common grown soft robots S-type Biped L-type Zigzag
  • 9. 9 Training curve 24 independent runs each condition We tried a more complex task: 3D soft robots locomotion Result 1 Feed forward Feed forward Feed forward Feed forward Feed forward Recurrent Recurrent Recurrent Recurrent Feed forward 2D Group 3D Group 100 population and 300 generation Common grown soft robots Roller Jumper Pull-Push Slider Jitter Crawler Slider Pull-Push L-Walker Jumper
  • 10. 10 Partial Summary Partial Summary for growth capacity of soft robots We gave growth capacity to 2D and 3D soft robots through neural cellular automata. An advantage of the presence of cellular memory (LSTM unit) was depend on tasks. A morphology (e.g. Tripod) that did not appear in previous studies using CPPN [Cheney+2013, GECCO] was acquired.
  • 11. 11 We investigate the ability of the soft robots to regenerate their body parts to recover from morphological damage. NCAs that are trained for locomotion failed regeneration. Result 2 Regrowing 10 steps from damaged morphology
  • 12. 12 Training for regeneration of soft robots through genetic algorithm Get phenotype (Morphology of soft robots) Similarity Evaluation between the original morphology Fitness value (Similarity) [Such+2017, arXiv preprint] Evolve weight of genotype (NCA) Grow soft robots through NCA from damaged morphology (10 steps) Method: Regeneration task Regrown Original Compare position (x, y, z) and type of voxel (Muscle, Tissue, Bone, Empty) for 9*9*9=729 voxels.
  • 13. 13 Soft robots are almost regrown through NCA. Result 2 Training curve 1 independent run 1000 population and 1000 generation 100% (729/729) is completely consistent with the original morphology.
  • 14. 14 Soft robots are partially recover their locomotion. Result 2 Tripod Travel distance 40.4 1.63 (3.6%) 20.3 (45%)
  • 15. 15 Summary Summary We gave growth capacity to soft robots through neural cellular automata. An advantage of the presence of cellular memory (LSTM unit) is depend on tasks in our experiments. A morphology (e.g. Tripod) that did not appear in previous studies using CPPN [Cheney+2013, GECCO] was acquired. We gave regenerative capacity to soft robots through neural cellular automata. This model requires two different neural cellular automata for growth and regeneration.
  • 16. 16 Discussion [Kriegman+2020, PNAS] Biological organisms can memorize a global morphology through bioelectric signaling network. [Levin+2017, Annual Review of Biomedical Engineering] Bioelectric signaling network can modify using extrinsic electrical stimuli. Hybrid soft robots may realize the regenerative ability of machines. XenoBod