Teaching of Locomotion for a Multi-legged Robot based on Sound Perception

Abstract

Sound perception function and vector field control are effectively combined for a multi-legged robot system to realize the simple and elegant locomotion in complex environment. The high level controller of the robot determines the desired body movement based on the sound perception, and sends a set of resultant weight parameters to all local controllers of homogenous leg modules. The local controllers then determine the leg strides based on the weighted summation of basis vector fields and quickly adjust the locomotion through local communications. Experimental results show the effectiveness of our approach.