AI Website MakerExisting approaches in vision-based control literature utilize information about the robot's end effector pose in the image to close the control loop. Consequentially, these approaches only assert control in the robot's task space. My skeleton-based approach enables us to control the entire shape of redundant robots using purely morphological information.
We utilize clothoid curves to parametrize the robot's shape in the image. Shape features of the clothoid are used as image features in an uncalibrated visual servoing scheme. The proposed approach does not require any proprioceptive sensor information from the robot.
AI Website Maker