Event-Triggered Pixel Model Predictive Control of Visual Servo Systems

In view of the large deviation between the starting Angle and the expected arrival Angle when the logistics transport robot performs the visual servo task, the feature points will often be lost in the field of vision due to the large attitude change of the robot during the task, resulting in the failure of the task. In this paper, an event-triggered model predictive control based on pixels is proposed, and a time-varying prediction model for mobile robot vision is established. By introducing a cost function that increases the direction of feature points, the error is reduced while the redundant motion of feature points is reduced, so as to control the feature points to remain in the field of vision. After the feature points are removed from the risk of loss, the optimization function is switched to reduce the time cost of optimization. Compared with ordinary MPC control, it can avoid the loss of feature points in the process of motion and reduce meaningless optimization operations.