Realisation of an adaptive real-time joint controller for a closely coupled two-arm manipulator system

A joint controller is used as a basis for a sensory-based, closely coupled cooperating two-arm manipulator system. Mechanisms for the synchronization of all joint controllers as well as for a close interaction between the joint control and the cartesian level have been realized. The control accuracy under high real-time demands has been improved with the help of the adaptation strategy of parameter scheduling. Control parameters for the scheduling that accord significant, linearized working points are extracted by an offline identification and optimization process. Therefore, motions of a PUMA260 manipulator characterizing the specified working points are recorded under control of an algorithm with manually determined parameters. A simplified model of the process is achieved by a recursive-least-squares mechanism, and the optimized parameters are extracted by using a heuristic method