Evolving Soft Robots

Built a custom 2D physics engine and genetic algorithm from scratch in Python to evolve soft-body robots capable of locomotion. Each robot is represented as a spring-mass network; the GA selects and mutates the fittest individuals across generations.

Over 100,000 generations, the best robots achieved a peak speed of 0.6 m/s. The "zoo" video shows the diversity of evolved body plans — from crawlers to hoppers — demonstrating the algorithm's ability to discover varied locomotion strategies.