HOMOGRAPHY-BASED VISUAL PREDICTIVE CONTROL OF TRACKED MOBILE ROBOT WITH FIELD-OF-VIEW CONSTRAINTS

Bingxi Jia and Shan Liu

Keywords

Homographybased, key frame strategy, field of view, nonholonomic constraints, model predictive control

Abstract

In this paper, a visual servo controller is proposed for a nonholonomic tracked mobile robot with field-of-view (FOV) constraints imposed by the vision system. The control scheme is directly expressed in terms of individual entries of the homography matrix, without the need of 3D reconstruction in conventional methods. To loosen the FOV constraints, key frame strategy is proposed, in which the homography is defined between the current image and the key frame image, instead of the target image. Thus the landmark can be invisible temporarily, and a better path in Cartesian space can be achieved. To deal with the nonholonomic constraints, a path planner is proposed to get a feasible trajectory, and model predictive control is used for trajectory tracking with physical constraints. Simulation results are provided to show the effectiveness of the proposed method and its advantages against conventional methods.

Important Links:



Go Back