An Optimal Planning Framework to Deploy Self-Reconfigurable Modular Robots

Abstract

Self-reconfiguration is a hard problem due to the high dimensionality of self-reconfigurable modular systems. Searchbased approaches offer complete and optimal solutions. However, naive search algorithms cannot directly solve self-reconfiguration tasks in reasonable time. In this letter, a transition model, a search heuristic, pruning strategies, and offline computations are proposed to advance performance of search-based self-reconfiguration methods. The targeted reconfiguration tasks include the reconfiguration of any distribution of modules, whether it is a complex structures or disconnected modules distributed in space. Hypothesized strategies are tested with conventional search methods to understand their contributions on complexity, optimality (minimum number of steps), completeness, and time efficiency. The whole framework is designed keeping hardware restrictions in mind and deployed on real Roombots hardware.