A Study on Hexapod Gait Adaptation by Enumerative Encoding and Particle Swarm Optimization

Abstract

Adapting to actuator failures is relevant robotic systems to continuously perform tasks such as exploration and mapping of unknown environments. This paper studies the adaptation ability to faulty hexapod legs by using an enumerative encoding scheme of the gaits locomotion strategy. The computational studies using the feasible set of leg failures through a physics-enabled simulation environment show (1) the feasibility of using the factoradic representation to explore feasible hexapod gait recoveries, and (2) the effectiveness of using an exploitative particle swarm-based heuristic to find feasible recovery strategies with minimal deviation to pre-defined locomotion commands under a small number of objective function evaluations. The results has the potential to further explore the enumerative encoding scheme and population-based optimization heuristics to render feasible hexapod gaits when actuators become malfunction on the field.