Evolutionary swarm robotics
← Back to homeI received the NSF GRFP in 2024 to study the impacts of different substrates and selection schemes on the evolution of swarms. This work is ongoing, with the first publication accepted at IEEE SSCI 2025.
We show that Markov Brains and Cartesian Genetic Programming can be used to control agents in swarms, alongside traditional neural networks. Swarms can even be composed of multiple types of controllers at once to facilitate specialization. Further work will expand this to more controller types, more complex tasks, and more selection schemes.
Made possible by MABE, with help from Clifford Bohm.
