A comprehensive account of how robots can evolve, learn, and adapt, bridging evolutionary computation, embodied intelligence, and the emerging frontier of physical AI. Accompanied by open-source software for experimentation, prototyping, and reproducible research.
If evolution can create intelligence, then artificial evolution can create artificial intelligence.
A population of robots is dropped into a remote forest. Bodies and behaviours evolve on-site to map terrain, monitor biodiversity, and adapt to conditions no human engineer ever anticipated.
Learn more →
A colony of self-evolving robots is sent to a hostile, uninhabitable planet. Generation by generation, they become better adapted to the environment until they can not only survive and reproduce, but also manipulate the world to prepare it for human habitation.
Learn more →
Three waves are reshaping AI: generative, agentic, and physical. Robot evolution draws on all of them at once; it generates morphologies and controllers, gives robots agency over their own behaviour, and is ultimately grounded in the physics of the real world.
Cognitive and creative capabilities now widely taken for granted. In robot evolution: the engine that proposes new morphologies and controllers.
Autonomous, goal-directed action by digital agents. In robot evolution: the robots have agency and autonomy, their behaviour determines their fitness.
Intelligence in physically embodied systems that interact with their surroundings. In robot evolution: robots 'live' in an environment where the laws of physics determine the results of the robots' actions. This holds even if the environment is only simulated.
For Bachelor, Master, and PhD students in AI, computer science, robotics, or related fields — an accessible entry point with worked examples.
For those looking to bring the latest techniques in generative AI, evolutionary algorithms, and machine learning to embodied, adaptive and intelligent systems.
For engineers and researchers building autonomous systems — a new perspective on design, adaptation, and physical intelligence.
Evolving robot populations as a controllable research platform to gain new insights into evolution and embodied behaviour.
A modular, extendable, open-source framework that lets you evolve and train robots — bodies, brains, and behaviour — in simulation and on real hardware.
Available from Springer in the Natural Computing Series. Hardcover, paperback, and ebook editions.