2D trifocal tensor based visual servo regulation of nonholonomic mobile robots

For a nonholonomic mobile robot system equipped with a fixed camera, we propose a two-dimensional trifocal tensor (2DTT) based visual servo regulation strategy in this paper. Firstly, the derivation of 2DTT is introduced and then estimated using feature based algorithm. Subsequently, a feedback linearizing position controller is designed to regulate the position errors of the robot, then, a proportion controller is designed to regulate the rotation error. The regulation task is implemented in the presence that the translation parameter and scene information are unknown. Simulation results are collected to investigate the feasibility of the proposed approach. Compared with existing methods, the proposed strategy makes full use of 3 images obtained in the servoing process, thus it is more robust to image noises than 2-images-based methods; the field of view constraint is largely alleviated because 3 points are enough to estimate the 2DTT; we can obtain exponentially convergent rate for closed-loop systems, and any pose estimation algorithms are not used.