Adaptive output stabilization of manipulators

This paper considers the position regulation problem for rigid robots for the case in which only joint position measurements are available, and proposes two adaptive controllers as solutions to this problem. The first controller is developed by assuming that the structure of the vector of gravity torques is known, but that the inertial parameters for the manipulator and payload are unknown; it is shown that this scheme ensures semiglobal stability and convergence of the position error to zero. Alternatively, the second adaptive strategy is derived under the assumption that no information is available concerning the manipulator model, and in this case it is shown that the controller provides uniform boundedness of all signals and exponential convergence of the position error to a set which can be made arbitrarily small. Experimental results are presented for a Zebra Zero manipulator and demonstrate that the proposed approach provides a simple and effective means of obtaining high performance position regulation.<<ETX>>