Evolving Self-Assembly in Autonomous Homogeneous Robots: Experiments with Two Physical Robots

Abstract

This research work illustrates an approach to the design of controllers for self-assembling robots in which the self-assembly is initiated and regulated by perceptual cues that are brought forth by the physical robots through their dynamical interactions. More specifically, we present a homogeneous control system that can achieve assembly between the modules (fully autonomous robots) of a mobile self-reconfigurable system without a priori introduced behavioural or morphological heterogeneities. The control system is an evolved dynamical neural network that directly controls all the actuators and causes the dynamic specialisation of the robots by allocating roles between them based solely on their interaction. Our results suggest that direct access to the orientations or intentions of the other agents is not a necessary condition for robot coordination: our robots coordinate without direct or explicit communication, contrary to what assumed by most research works in collective robotics. Finally, we show that our evolved controllers prove to be successful when tested on a real hardware platform, the swarm-bot. The performance of our evolved neuro-controllers is similar to the one achieved by existing modular or behaviour-based approaches, owing to the effect of an emergent recovery mechanism that was not foreseen during the evolutionary simulation. However, contrary to other approaches, our system proved to be robust against changes in the experimental setup, because of the reduction in the number of user-defined preconditions for robot self-assembly.