Evolution of Neuro-Controllers for Robots' Alignment using Local Communication

In this paper, we use artificial evolution to design homogeneous neural network controller for groups of robots required to align. Aligning refers to the process by which the robots managed to head towards a common arbitrary and autonomously chosen direction starting from initial randomly chosen orientations. The cooperative interactions among robots require local communications that are physically implemented using infrared signalling. We study the performance of the evolved controllers, both in simulation and in reality for different group sizes. In addition, we analyze the most successful communication strategy developed using artificial evolution.