An evolutionary swarm self-assembly robot: From concept to prototype

This paper presents algorithm design and prototype experiments on an evolutionary swarm self-assembly robot (Sambot). Multiple Sambots can be self-assembled to form a robotic structure further to generate the whole-body locomotion. For whole-body locomotion, a unity control method is more difficult to design than for ordinary robots because self-assembly system can form various configurations. This study takes function, morphology and the corresponding control system into consideration using an evolutionary idea. A unity CPG model as well as unity configuration representation is proposed to generate locomotive motions for self-assembly systems and improve self-adaptation according to any configuration. Based on it, the simulation for a linear configuration is done and besides the hardware experiments from self-assembly to locomotion are implemented on Sambot platform.