Gait Planning and Control for a Hexapod Robot on Uneven Terrain Based on Markov Decision Process

Abstract

Gait planning of hexapod robots the application prospect of which is wide in various fields is very complicated because of there abundant gaits and sophisticated limb structure. By establishing a discrete gait model and a stable position state space of the robot, the complicated problem is transformed into an optimal sequence decision-making problem based on the stability margin. According to the Markov decision process, the problem can be solved. The Monte Carlo method is used to optimize gait strategy. The average stability margin and convergence rate in gait can be raised greatly under this method in which the state transition probability does not need to set. When a robot encounters bumps and depressions, it is easy to lose stability. In this paper, a method of adjusting the time of foot supporting and swing is proposed to solve the problem. Experiments in the Webots environment show that the method of planning and control can quickly plan the optimal gait and adjust in real time the gait for uneven terrain, also improving the walking stability of Hexapod robot.