Trajectory Tracking Based On Model Pre Dictive Control For Soccer Robot

Under the background of RoboCup Middle Size League soccer robot competition, an efficient approach based on Bezier curve and Voronoi diagram for path planning is proposed in the thesis, and then the trajectory tracking problem of the omnidirectional mobile robot is researched as the main purpose.Path planning is one of the most significant techniques for soccer robots to win the game. The obstacle avoiding methods and path planning algorithms which are widely used by RoboCup Middle Size League soccer robots are reviewed firstly, and the draw-backs of these methods are pointed out, and then the feasibility of the path planning al-gorithm based on the global world model is analyzed. According to the positions of ob-stacles and the characteristic points of the competition field, the Voronoi diagram can be generated, and then the reasonable control points are chosen to build a smooth Bezier curve which could enable the soccer robot pass obstacle regions safely and shoot the goal directly. Finally the geometric path is adjusted to be a reasonable trajectory with defined acceleration and velocity.To improve the dynamic performance of trajectory tracking, hierarchical control based on kinematic model is employed, and then based on the linear full dynamic error model, MPC is developed to design the control law to satisfy the kinematics constraints, and the Laguerre Function is used in MPC design to guarantee good dynamic perform-ance in trajectory tracking and reduce the needed computation time.in comparison with the nonlinear error model, the controller based on the lin-earized models could only guarantee the local stability and the local controllable prop-erty is lost when the linear speed and angular speed converge to zero. To solve these problems, nonlinear model predictive control based on the nonlinear error model is re-searched, the control stability is guarranteed by adding a terminal state penalty to cost function and constraining the terminal state to a terminal state region.The experimental results show that the proposed path planning algorithm and tra-jectory tracking algorithm are effective, and the motion performance of RoboCup Mid-dle Size League soccer robots can be improved greatly.