Generation Method of Quadrupedal Gait Based on Human Feeling for Animal Type Robot

Abstract

Characterizing animal movement appropriately is important in pet robot design and Robot-Assisted Therapy (RAT). In studying how to generate animal gaits in quadruped robots, we used the canine AIBO robot. We began by optimizing a single leg path outputting propulsion efficiently and imitating canine walking using a genetic algorithm and a zoological basis. We administered a questionnaire to help determine subjective feelings in observers for choosing the optimum AIBO gait. We corrected minor deviations in joint parameters to stabilize walking on the ground.