2P1-C10 Circular Path Estimation of a Rotating Four-Legged Robot Using a Hybrid Genetic Algorithm LSM

Abstract

There exists an interest for a new type of vehicle which inherits both advantages of wheeled and legged locomotion. Control of such robots is not a trivial problem since we must find algorithms and strategies for utilizing the type and exhibiting complex interactions with environment. This paper presents the step toward developing a C++ object oriented based simulator for a rotating four-legged mobile robot called PEOPLER which can behave as a four wheeled or four legged robot on demand. We propose a new hybrid method for estimating circular radius of the legged locomotion path on a flat terrain. We call the hybrid estimation method GALSM as an expression of combining genetic algorithm (GA) and the well known least square method (LSM). Results of the estimation showed that the GALSM is promising in steering of the robot.