No abstract

This paper lays out a framework to model the kinematics and dynamics of a rigid spacecraft-mounted multibody robotic system. The framework is based on dual quaternion algebra, which combines rotational and translational information in a compact representation. Based on a Newton-Euler formulation, the proposed framework sets up a system of equations in which the dual accelerations of each of the bodies and the reaction wrenches at the joints are the unknowns. Five different joint types are considered in this framework via simple changes in certain mapping matrices that correspond to the joint variables. This differs from previous approaches that require the addition of extra terms that are joint-type dependent, and which decouple the rotational and translational dynamics.

The problem of estimating the mass properties of a spacecraft while tracking a 6-DOF reference is addressed using dual quaternions. Dual quaternions provide a position and attitude (pose) representation, which has proven to be advantageous over other, more conventional, parameterizations. An adaptive controller for 6-DOF tracking is proposed using concepts from the concurrent learning framework. The latter is a recently proposed methodology to incorporate current and recorded system data from measurements into the update of an adaptive controller's parameters. Asymptotic convergence of the parameters is ensured through an easily verifiable rank condition of the matrix formed from a finite set of collected data, contrary to the rather stringent, but more common requirement of persistency of excitation. Simulation results for the tracking of a non-persistently exciting, 6-DOF reference are provided and compared to the baseline adaptive controller.

No abstract

No abstract