Evo-Bots: A Simple, Stochastic Approach to Self-assembling Artificial Organisms

Abstract

This paper describes an alternative path towards artificial life—one by which simple modular robots with novel hybrid motion control are used to represent artificial organisms. We outline conceptually how such a system would work, and present a partial hardware implementation. The hardware, a set of self-reconfigurable modules called the evo-bots, operates on an air table. The modules use a stop-start anchor mechanism to either rest or move. In the latter case, they undergo semi-random motion. The modules can search for, harvest and exchange energy. In addition, they can self-assemble, and thereby form compound structures. Six prototypes of the evo-bot modules were built. We experimentally demonstrate their key functions, namely hybrid motion control, energy harvesting and sharing, and simple structure formation.