A multi-stage minimum time adaptive control of robot manipulators

This paper introduces a new approach to multi stage switching control based on the dynamic model of robot manipulators. In the first stage, a Minimum Time Preparation control (MTP) is applied. It is shown that the time-optimal control of a six degrees of freedom manipulator is merely of bang-bang type in the space of the generalized coordinates, with singularities in the constant speed zone. The second stage is an Adaptive Approach (AA) feedback control with model adaptation. This is applied to ensure proper approach and accuracy in attaining the desired target in the task space. A Neuro-Fuzzy Switching (NFS) learning system is built up to characterize the best switching instant time. This approach was applied for an assembly robot of pick and place task. Simulation results showing effectiveness of the proposed method are given.