Publications Details

Visual servoing using statistical pressure snakes

Schaub, Hanspeter

A nonlinear visual servoing steering law is presented which is used to align a camera view with a visual target. A full color version of statistical pressure snakes is used to identify and track the target with a series of video frames. The nonlinear steering law provides camera-frame centric speed commands to a velocity based servo sub-system. To avoid saturating the subsystem, the commanded speeds are smoothly limited to remain within a finite range. Analytical error analysis is also provided illustrating how the two control gains contribute to the stiffness of the control. The algorithm is demonstrated on a pan and tilt camera system. The control law is able to smoothly realign the camera to point at the target.