Intelligent Robotic System with Fuzzy Learning Controller and 3D Stereo Vision

It is well known that robotic manipulators are highly nonlinear coupling dynamic systems. It is difficult to establish an appropriate mathematical model for designing a model-based controller. The model-free feature of fuzzy logic control strategy was employed to design robotic motion controller. However, there is no guide rule for designing the fuzzy rule bank and parameters, it still needs time consuming trial-and-error work for rules bank and fuzzy parameters adjustment. We had developed self-organizing fuzzy control and adaptive fuzzy sliding mode control two intelligent learning mechanism for solving this implementation problem. In addition, a low cost stereo vision system is developed on chip processor. It can be integrated into robotic system for executing visual servo robotic motion control purpose. Both systems can be constructed on Nios II SOPC developing board with ALTERA FPGA chip to manipulate a retrofitted Mitsubishi robotic system. The 3-D position information between the target and stereo vision system was extracted by stereo vision algorithm first. Then, the relative motion between the robotic end-effector and the target can be planned to guide robot arm to catch the object. The self-organizing fuzzy control and fuzzy sliding mode control algorithms are employed to monitor the trajectory motion of each joint. The experimental results show that this visual servo robotic system can track and catch a moving target in 3D space and execute some interaction functions with player. Key-Words: visual servo, robotic system, self-organizing fuzzy control, fuzzy sliding mode control, FPGA chip