Image-Based Regulation of Mobile Robots Without Pose Measurements

摘要

This letter presents an image-based regulation control of nonholonomic mobile robots using a single onboard camera. State-of-the-art approaches depend on pose reconstruction or decomposition of homography for the position-based visual servo. A new image-based switching control scheme with a hierarchical structure is developed. The proposed approach does not depend on pose estimations and works on both planar and non-planar scenes. The dynamics is newly formulated in terms of invariant features to rotation, and a specific invariant feature introduced by Tahri <i>et al.,</i> is implemented in the practice. The applied features can be easily calculated according to at least three feature points. The closed-loop system is proved to be exponentially stable in the Lyapunov sense. The performance is validated through 90 experimental trials in three different environments without artificial markers. Results show the ability of regulation using direct image measurements for auto-charging, picking up goods, and home service. All of the approaches are performed onboard the vehicle.