A neurofuzzy model-based compliance controller with application to a telerobot system

This paper proposes a new design method for a compliance controller in a telerobot system which interacts with an environment having unknown stiffness. A neurofuzzy compliance model (NFCM)-based control proposed here is a control scheme designed to determine automatically the suitable compliance for a given task. An NFCM, composed of a fuzzy logic controller and a rule-learning mechanism, is used as a compliance controller. The fuzzy logic controller receives contact forces as inputs and generates corresponding corrective motions as outputs. The rule-learning mechanism, composed of two neurons, trains the rule base of the fuzzy logic controller until the given task is successfully performed by using a reinforcement learning algorithm. The scheme does not require any prior knowledge on the slave arm dynamics, slave arm controller or the environment, and thus it can be easily applied to the compliance control of telerobot systems, The effectiveness of,the proposed scheme is verified through a series of experiments using a laboratory-made telerobot system.