Model Predictive based Trajectory Tracking Control of Quadruped Robot

Abstract

Quadruped robots are four legged robots which has broad applications in complex situations such as military investigation, inspection, planet exploration and so on. The problems with quadruped robots are difficulties to keep their balance during standing and movements. These have been studied and somehow fixed by the researchers and now the difficulty is improving the performance of the robots. So, for the improvement of the trunk centre of mass trajectory tracking accuracy plus to provide proper leg angles behavior for the trajectory of the leg, an efficient control method is required. This thesis presents model predictive based tracking control for the trajectory of quadruped robot. The method proposed involves developing a dynamic model for the quadruped robot to determine the ideal leg contact force that tracks the center-of-mass trajectory. Both linear and circular trajectory was given for the trunk system as a reference input so that it will track it by the provision of proper controller design parameters. The basic parameters of the controllers are tuned with help of a global metaheuristic optimization algorithm called PSO method. As simulation result shows in linear trajectory the steady state error is 9.7702e-05m or 0.0097702%. In general, the proposed controller have met the objective of the study which is important to make the robot perform different given tasks with low time period and with high efficiency.