This paper gives an overview of the nonlinear adaptive control work that was completed at the Stanford University Aerospace Robotics Laboratory (ARL) in December 1992. A new task-space adaptive control framework was developed that is able to provide continuously full adaptation capability to complex robot systems in all modes of operation. This framework consists of an inverse-dynamics adaptation algorithm that has been generalized beyond simple joint or endpoint control, a new system modelling technique to simplify the generation of a system model to ease greatly the implementation of the adaptive control algorithm, and the development of the task-space concept to allow operators to specify a robot's task, which can include payload positions, endpoint positions, and joint configurations as subsets. The task-space adaptive control framework has been experimentally demonstrated on the ARL Multi-Manipulator, Free-Flying Space Robot performing capture and manipulation of free-floating objects with unknown inertial properties-without requiring human assistance.
展开▼