Dynamic simulation of robot manipulators at high sampling rates

The authors introduce parallel-processing concepts to the problem of the real-time dynamic simulation of a robot manipulator. A heuristic load-balancing procedure is used to distribute the dynamic equations of a robot arm over a network of parallel-processors. The dynamics are divided into different subtasks. Each inter-related group of subtasks is assigned to one processor. This helps in minimizing the communication time between different processors. In this case, each processor will spend its time on useful computations and number crunching, leading to better utilization of the resources. The efficiency of this approach lies in its simplicity, generality, and cost-effectiveness. The method is demonstrated by a case study