Navigability of multi-legged robots

Abstract

This paper addresses the improvement of navigability for a six-legged robot through the development of a simple method for measuring heading and drift errors. More specifically, the navigation scheme utilizes both a magnetic compass and landmark navigation to correct these errors with every step, hence limiting error propagation. The approach is aimed at operation in unknown environments. Elaborate data processing in the control algorithm is avoided by using modular sensors capable of processing their own inputs. The robot controller uses the well-known tripod gait, implemented here using the hexapod inverse kinematics. Experimental results show significant improvements in the hexapod's navigability for turning and walking maneuvers.