EYE-IN-HAND ROBOTIC TASKS IN UNCALIBRATED ENVIRONMENTS

Flexible operation of a robotic agent in an uncalibrated environment r equires the ability to recover unknown or partially known parameters o f the workspace through sensing. Of the sensors available to a robotic agent, visual sensors provide information that is richer and more com plete than other sensors. In this paper we present robust techniques f or the derivation of depth from feature points on a target's surface a nd for the accurate and high-speed tracking of moving targets. We use these techniques in a system that operates with little or no a priori knowledge of object-and camera-related parameters to robustly determin e such object-related parameters as velocity and depth. Such determina tion of extrinsic environmental parameters is essential for performing higher level tasks such as inspection, exploration, tracking, graspin g, and collision-free motion planning. For both applications, we use t he Minnesota robotic visual tracker (MRVT) (a single visual sensor mou nted on the end-effector of a robotic manipulator combined with a real -time vision system) to automatically select feature points on surface s, to derive an estimate of the environmental parameter in question, a nd to supply a control vector based upon these estimates to guide the manipulator.