Mechanical design and computational aspects for locomotion and reconfiguration of the ModRED Modular Robot

In this paper we describe the mechanical construction and AI-based planning techniques for the locomotion and reconfiguration of a modular self-reconfigurable robot (MSR) called ModRED (Modular Robot for Exploration and Discovery). ModRED is a highly dexterous, chain-type MSR with 4 degrees of freedom. It can maneuver in tight spaces and is suitable for autonomous locomotion over unstructured and uneven terrain such those encountered in extraterrestrial environments like the surface of the Moon or Mars, or, in environments that are difficult to navigate for humans like the inside of a volcanic crater. ModRED uses gait tables for navigating in a fixed configuration. We have developed a fuzzy logic control based approach for dynamically adapting ModRED's gait without changing its configuration, if its goal changes or if its motion is impeded by an obstacle. We also describe our research on reconfiguration planning in ModRED using a coalition game theory based technique that allows the modules to dynamically reconfigure into a new shape while reducing the time and cost expended to achieve the new configuration.