Integrated Adaptive Control of Space Manipulators

Abstract

Indirect adaptive control of space robot manipulators when maneuvering payloads with imperfectly known mechanical parameters is considered. The objective is to estimate system mechanical parameters during the maneuver itself. The system bodies are rigid and are concatenated by ideal, rotational joints. A new adaptive control law is developed that is applicable to general, rigid, multibody systems. It is based on the reinterpretation ofthesystemdynamicequationsasameasurementequation.Theadaptivecontrollawisoftheintegratedtype;that is, the estimator part is used to estimate the integrated ine uence of the system mechanical parameters rather than the parameters themselves. The system equations of motion, control, and estimation are presented, and a formal solution is given. The parameter estimation process and the control law are analyzed separately. For constant generalized parameters, it is shown that the control-system output error is globally asymptotically stable and that theparameter erroralso will convergeto zero if theexternal command input satise es a certain sufe cient excitation condition. A numerical simulation of a planar free-e oating spacecraft with a two-degree-of-freedom robotic arm handling various payloads is presented. I. Introduction A LONG with the rapid development of the space industry during the last few decades, the need to lessen the actual human participation in space systems in order to save human efforts and avoid hazards has become clear. Autonomous robotic systems are about to become part of reality in space missions. Considerable research has been directed to some primary functions of robots in space applications, 1i7 as well as to the related technical issues such as kinematics, dynamics, and control. 7i26 Robot manipulators carrying out extravehicular operations in spacetypicallyhandlepayloadswithimperfectlyknownmechanical properties. Manipulator parameter and structural uncertainties, inaccuracies, modeled and unmodeled disturbances, and various couplings and nonlinearities not included in the model also appear in real tasks. Online estimation of those parameters and signals improves the quality of the maneuvers. To achieve this objective, recourse is taken to indirect adaptive control, i.e., the system parameters used in the control law are adjusted online, during execution of the desired maneuver. Analyzing and simulating the existing manipulator (indirect) adaptive control methods 27i41 showed that, even for the simplest case of a one-link cone guration, a number of algorithmical and numerical problems are still open. 42 When one considers generalization to the multilink case, these problems become even harder, and many more analytical, algorithmical, computational, and numerical ones appear. The most critical is probably the determination of the regressor matrix, a task too demanding for high-order systems such as a multiarmed space robot. To avoid the need to generate the regressor matrix, a new adaptive control algorithm, the Integrated Adaptive Control, has been presented 42 to control multibody space robots. The algorithm is basically an indirect adaptive control method, in which the novelty is in the adaptive parameters dee nition. The generalized parameter matrices and vectors, which represent the integrated, time-varying effect of all unknown as well as supposedly known parameters, are directly estimated and tracked, rather than the basic mechanical parameters themselves as in all existing algorithms. This is achieved by reinterpreting the system dynamic equation as a linear timevarying measurement equation, in which those generalized parameters are time-varying parameters or state variables to be estimated. The generalized parameters vary slowly with time as applicable in