Visually Servoed Tracking with Redundant Degrees-of-Freedom

Presented is a partitioned controller for visually servoed tracking of objects using both image and the kinematic data. The resulting design shows remarkably similar performance to vision systems in biology. For example, people coordinate their eyes, head and body when tracking a moving target. Likewise, this controller couples a robot's degrees-of-freedom to point a camera at a moving target. Tracking experiments showcase the design on a robotic system featuring two revolute joints to mimic head-eye motions. Keywords—Biomimetic, computer vision, redundant degrees-offreedom and visual-servoing.