Real-time Minimum Jerk Optimal Trajectory Synthesis and Tracking for Ground Vehicle Applications

Abstract

Research on wheeled mobile robots has been an active research field for several decades with focus on the problems of stability, maneuverability and control. The trajectory generation for wheeled mobile robots is usually handled by considering a smooth function satisfying the boundary conditions. Even though this problem is well posed for several function classes, the real issue occurs in generating a feasible trajectory for the robot which takes into account the constraints of the system. Another problem from the control system perspective is, whether the derived controller is stable for the entire trajectory span to track the given trajectory. The purpose of this thesis is to provide a different approach for trajectory generation and control of an autonomous ground vehicle. The presented trajectory generation technique takes into account the acceleration constraint of the system by performing an optimization routine in-order to obtain the final time of the trajectory for a given waypoint. Then, the problem of tracking the desired trajectory is handled by using a nonlinear backstepping control law which takes into account the non-holonomic constraints of the robot. The overall experimental setup is based on the cyber-physical system architecture by separating the rover and the ground station, which handles all the necessary computation. The time delays associated with this kind of system architecture is characterized and presented. The presented trajectory generation and control techniques are experimentally verified by using the cyber-physical system architecture on a real mobile robot equipped with GPS, IMU, Wheel Encoder and LiDAR sensors.