Uncalibrated image-based visual servoing approach for translational trajectory tracking with an uncertain robot manipulator

This work presents an adaptive image-based visual servoing approach for uncertain robot manipulators using an eye-to-hand camera configuration, which does not require any calibration procedure and image velocity measurement. A monocular camera provides visual feedback for the robot's joint control, allowing successful tracking of the translational trajectory, prescribed in image space, for a given target object attached to the robot end-effector. As our contribution, an indirect adaptive control approach is proposed to deal with the uncertain parameters of the camera-robot system and any camera misalignment concerning the robot frame. The indirect adaptive visual servoing approach is then combined with a direct adaptive motion controller using a cascade control framework to consider the uncertain robot dynamics in our stand-alone robot vision system. Lastly, an adaptive model-based state observer is designed to avoid the need for measuring the image feature velocity in the interconnected control loops. The Lyapunov stability theory and the passivity paradigm are synergized to demonstrate the stability properties of the overall closed-loop system. Experimental results, obtained with a 4-DoF robot manipulator carrying out translational trajectory tracking tasks using a commercial webcam, illustrate the performance and effectiveness of the proposed adaptive control methodology.

Automated summary

This work presents an adaptive image-based visual servoing approach for uncertain robot manipulators using an eye-to-hand camera configuration, which does not require any calibration procedure and image velocity measurement. By combining the indirect adaptive visual servoing approach with a direct adaptive motion controller, the uncertain robot dynamics are considered, leading to a stable robot vision system.