An intelligent controller for robot contact and non-contact task control

An architecture for intelligent control of a robot that interacts with a dynamically changing environment is proposed. The problems that are encountered in the implementations of low-level controllers for such robotics applications and their inherent limitations are outlined. An intelligent monitor is proposed to overcome these problems. Since this intelligent interface can change the desired trajectory and thereby affect the stability of the low-level controller, a stability analysis of the integrated controller and the interface is provided. This conceptual architecture is an attempt to emulate humans in employing common-sense reasoning in dynamic decision making and control. It is strongly believed that this approach will enable the controller to generate control actions in a more 'humanlike' manner than the current implementations of learning control.<
>