Behaviour Subdivision and Generalization of Rules in Rule-Based Control of the ATRON Self-Reconfigurable Robot

The complexity of control is a major problem in self-reconfigurable robots. In this paper we present a method for reducing the complexity of controller design by dividing a global behaviour into sub-behaviours realized by sets of rules dependent on the local configuration. The coordination of sub-behaviours is realized by behaviour arbitrators based on simple distributed signals. Furthermore, the generalization of simple rules into fewer, but more general rules, resembling even smaller subbehaviours, is investigated in terms of changes of the global behaviour due to the introduction of overgeneral rules. It is shown by example that more general rules perform better, in terms of speed of the global behaviour, than simpler rules. The proposed methods are evaluated by constructing a cluster flow behaviour for a large group of ATRON modules.