Evolutionary and distributed computation methods for controlling embodied agents

The goal of this research project is to investigate the realm of evolutionary and distributed computation techniques for the automatic synthesis or adaptation of controllers for (possibly simulated) embodied agents which interact with an environment with the purpose of performing some predefined task. More in detail, the recruited researcher will design, develop, and experimentally assess one or more techniques based on evolutionary computation or distributed computing, mainly focusing on cellular automata, for the control of simulated robotic agents, especially the modular ones. The proposed techniques will be compared against the state-of-the-art in terms of effectiveness, efficiency, and capability to generate diverse solutions for a given problem, so that those solutions are overall robust to variations of problem features. Moreover, the generalization ability and the degree to which other evolutionary techniques could be applied will be investigated.