Obstacle Overcoming Gait Design for Quadruped Robot with Vision and Tactile Sensing Feedback

Abstract

We propose an obstacle overcoming gait method that combines vision, tactile sensation, trajectory planning and gait selection for a quadruped robot when the robot encounters obstacles of different heights on irregular terrain. OpenCV is used to recognize the regular and irregular terrain images so that the robot is able to perform trotting and walking gaits on these terrains, respectively. A gait planning to overcoming obstacles is performed by calculating the position of the feet through forward kinematics combining the tactile sensor data in real time. The foot trajectory planning is realized with constrains by position, speed, acceleration and inverse kinematics. The designed obstacle overcoming gait is verified by Webots simulation and experiment on a prototype quadruped robot. The results prove the effectiveness and stability of the proposed method.