Biological “blind-tracking” task through artificial intelligence

The objective of this study was to identify a biologically inspired intelligent control structure that can solve the robot's dynamic hybrid control problem. This control structure was investigated as “blind-tracking” tasks for a human operator, in view of equivalent problem of force and position control for a robotic hand. To perform the blind-tracking task, three levels of coordination were required, namely; learning level, skill level and adaptation level. These levels are representing the human cognitive agents. Compared with artificial intelligence the learning level assigned to position control subspace with Neural Network, the skill level assigned to force control subspace with PI Fuzzy Logic, and adaptation level assigned to tool's contact surface identification with Fuzzy rules. The proposed control process was justified through the different type of human operator's performance. Naturally the human operators can be indexed from their past experiences, skills, and, guesses through their age and work habit. Thus the elements of the intelligent control model can be validated and established. The model was tested in the real robot.