Evolutionary Robotics

Evolutionary robotics studies how populations of candidate robots can be improved through selection and variation, often optimizing a robot’s body and controller together. Common approaches include genetic algorithms, evolutionary strategies, quality-diversity methods such as MAP-Elites, and coevolution, sometimes combined with reinforcement learning or neural-network controllers. Researchers test these methods on simulated soft-robot locomotion, manipulation, multi-robot coordination, and physical robots, with growing attention to transferring designs from simulation to hardware. A central tension is that encouraging diverse or novel designs can make useful traits harder to preserve, while learned controllers may be costly to evaluate and unreliable when bodies or environments change.