D.-J. Kim, R. Lovelett, Z. Wang, and A. Behal (USA)
Visual servoing, Estimation, Dynamicallychanging feature
We propose a region-based switching scheme for visual servoing to resolve two visibility issues for practical vision systems, namely, non constant features and limited Field-of-View (FOV) of the camera. For purposes of keeping target object in the camera’s FOV, region-based decision criteria are utilized to design a switching controller. Specifically, this controller utilizes feedback from a dynamically changing feature set to switch between a hybrid visual servoing scheme that generates 6-DOF motion in the end-effector and a 3-DOF scheme that generates only translational motion. Through a reference image as well as feature feedback available in the neighborhood of the missing data, an affine transformation is utilized to reconstruct missing features - these are utilized to estimate the largest convex hull for the target object. This augmented feature set is then utilized to switch between the controllers. The performance of the algorithm is illustrated via experiments conducted on the 6-DOF assistive robotic system.
Important Links:
Go Back
