Research on Path Tracking Predictive Control for Vision-Guided AGV

A predictive controller based on global gradient descent algorithm is designed for the problem of the path tracking of the Vision-guided AGV with nonholonomic constraints characteristics. The predictive controller can generates a set of control sequence which could eliminate all errors simultaneously under the constraints of control input and the constraints of state. In the time of solving optimization problems, a set of initial control sequence is generated based on the principle of gradient descent, and then optimizes the initial sequence from a global perspective with improved gradient descent algorithm which takes the entire control sequences as optimization goals. It can solve the affection of various constraints to AGV control effect. This algorithm is simple, targed, and the control process is fast and smooth, and meet the actual control needs.