Robots! (5/5): Self-assembling robots and the potential of artificial evolution | Emma Hart
Summary
Computer scientist Emma Hart presents a radical approach to robotics using artificial evolution to create machines that adapt to unknown environments, such as deep ocean trenches or distant planets. Instead of pre-designing robots, this technology sends a robot-producing system that builds and evolves physical robots in situ, mimicking biological natural selection through digital DNA, 3D printing, and automated assembly. By combining physical robots with virtual simulations, the system accelerates the breeding of optimized designs that can handle complex tasks like drilling or climbing. Har…
Topics discussed
Sponsor: Capital One's multi-agent AI for car shopping
Sponsor: Planet Visionaries podcast with Rolex
Sponsor: Olympic Hospitality for LA 2028
Sponsor: LA 28 Hospitality package details
Introduction: The challenge of designing robots for unknown environments
Concept: Robots that design, build, and evolve themselves
Use cases: Nuclear waste, deep sea mining, and asteroid exploration
Inspiration: Biological evolution and artificial evolution
Distinction: Biomimicry vs. harnessing evolutionary creativity
History: Artificial evolution in design and the need for robot brains
Mechanism: The three ingredients of evolution and digital DNA
Fabrication: 3D printing and automated assembly of robot bodies
Selection: Scoring performance and breeding new generations
Optimization: Using simulation to speed up the evolutionary process
Hybrid breeding: Combining physical and virtual robot populations
Applications: Exoplanetary exploration and sustainable robotic design
Sponsor: Capital One's Chat Concierge AI
Sponsor: Granger for shipping and industrial supplies
Sponsor: Perk platform for travel and spend management
Listen ad-free on Castria