Fast parallel preconditioned conjugate gradient algorithms for robot manipulator dynamics simulation (Q1335017)

The manipulator dynamics simulation problem, also called forward dynamics, concerns the determination of the motion of the mechanical systems resulting from the application of a set of joint forces or torques, respectively. The motivation of devising fast algorithms for forward dynamics computation stems from applications which require extensive off-line simulation capability (e.g. trajectory planning and design) as well as applications which require real time dynamic simulation capability. Authors show with their contribution that, by exploiting the inherent structure of the forward dynamics problem, not only suitable preconditioners can be devised to accelerate the iterations but also, based on the choice of preconditioners, a modified dynamic formulation can be used to speed up both serial and parallel computation of each iteration. The implementation of the parallel algorithms on two interconnected processor arrays is discussed in detail, and their computation and communication complexities are analyzed. By means of an example, the effectiveness of the preconditioners is illustrated. The paper is well readable and will be of interest to specialists in either robot dynamics or multibody system dynamics.