Uncalibrated Vision-Based Control of Two-Wheeled Mobile Robots

摘要

Abstract An quasi-Newton method with Jacobian estimation is used to control a mobile robot utilizing visual feedback. The method is uncalibrated, requiring no camera calibration or known robot kinematics. Given a proper task configuration, the robot can be controlled such that it follows a moving target. This paper investigates the appropriate task configurations that result in a controllable system.