ABSTRACT
Conventional robotic systems are limited to operating in highly structured environments, and considerable efforts must be expended to compensate for their limited accuracy. This is because conventional robotic manipulators operate on an open kinematic chain for placing and operating a tool (or endeffector) with respect to a workpiece. This is done by joint-to-joint kinematic transformations in order to define the
pose
(position and orientation) of the endpoint with respect to a fixed-world coordinate frame. Similarly, the workpiece must be placed accurately with respect to the same coordinate frame. Any uncertainty or error about the pose of the endpoint or workpiece would lead to task failure. Potential sources (such as gear backlashes, bending of the links, joints slippage, poor fixturing) would contribute to errors in the endpoint or workpiece poses. Therefore, considerable effort and cost are expended to overcome the above issues, for example, to design and manufacture special-purpose end-effectors, jigs, and fixtures. Consequently, due to the need for an accurate world model, the cost of changing the robot task would be quite high.
