Bent leg walking gait design for humanoid robotic child-iCub based on key state switching control

Abstract

Inspired by the observation that the 8-11 month old infant can learn to walk under the help of walker, we design a walker for the humanoid robotic child-iCub. A new gait - bent leg walking - is adopted to solve the problem that during walking the height between the waist of iCub and the ground should remain unchanged because the robot is rigidly attached to the walker. Based on the leg kinematics model, six key states and sub gaits switching between the adjacent key states are defined for iCub realizing bent leg walking under a given height. Within YARP framework, a controller is developed for realizing the algorithm of the proposed bent leg walking gait. Through inputting command from a YARP port, the controller can generate corresponding motor commands for iCub. The experimental results show that the proposed gait can control iCub Simulator to realize bent leg walking under the help of walker. This research provided a new gait for bipedal robot walking while keeping the height of the upper body unchanged, and a platform for the study of autonomous walking development, available in open source.