A Consideration of Dynamic Control of Robotics Manipulators Using MRAS
model-referenced adaptive control system
polar coordinate type manipulator
The effectiveness of the model-referenced adaptive control system (MRAS) developed by Dubowsky et al. has been investigated from the simulation analysis of the systems using such dynamic models as a simple linear 2nd-order type manipulator, a single degree-of-freedom linkage type one, and a 2 -degree-of freedom polar coordinate type one. When the tuning gains for the adaptive control of manipulator dynamics, Kp and Kv, are initially set at their ideal values, good performance of MRAS can be obtained over a wide range of configulational motions and inertia paylords, whether there is the nonlinear complex and coupled nature of the manipulator dynamics or not. Simulation results show that the practical limitation of the MRAS algorithmus in high speed manipulator applications lies in P (normalized frequency)≦2, and also that the offset due to the gravity effect can be improved by adding the I-action parallel to the Kp (position gain). The applications of the MRAS concept to so called automatic-tuning process control can be sufficiently expected in the sense that good settings of the parameters (feedback gains) are possible without the complex identification system about processes.
Departmental Bulletin Paper