The Arm/Hand integrated telerobotic system based on sensory information

This paper describes the Arm/Hand integrated telerobotic system and its multisensor integration and fusion. The characteristic of intelligent robotic system and the role of sensory data fusion in different abstract level are analyzed. A new flexible control structure is proposed to integrate different types of sensors. Neural network is used to obtain more abstract information from force and torque sensors, and knowledge based multisensor fusion method is developed to fuse the sensory information which is not only directly acquired from physical sensors but also from the middle fusion results in different fusion process level to make full use of the information hidden in the combination of fused information. Real time reasoning method and the integration of qualitative and quantitative information are also proposed in this paper. In addition, predictive simulation display is overlaid on actual live images with high fidelity to compensate for model discrepancies. With the multisensor integration and fusion, the teleoperated arm/hand integrated robotic system can perform some dexterous tasks such as screwing valve, peg in hole, and grasping a cup of water and pushing it to another cup.