Integrating fuzzy control of the dexterous National Taiwan University (NTU) hand

Abstract

Compared to traditional tendon-driven robots, the mechanism design of the National Taiwan University hand has an uncoupled configuration and is compact in size. In this paper, a specially designed compact control system, which is embedded into the NTU hand to satisfy the limited space, is developed to perform the control of the five-finger robot with 17 degrees of freedom (DOF). There are total 17 actuators including transmission mechanism, 17 potentiometers, and 18 tactile sensor pads to be integrated. Those 17 actuators should be controlled simultaneously by integrating 17 position sensing signals and 18 tactile signals. Multi-loop position control is further complicated by multi-loop force control as well as sensor integration. The proposed control system distributes the computation load into several modules and utilizes the DSP chip. It takes advantage of the fuzzy control by introducing the knowledge of human and sensor fusion with the finger joint and tactility. Using the communication function of the control system, the knowledge bases can be loaded for high level computation and modified during run time.