Optimal code generation for serial and parallel processing ofrobot-arm control computation

摘要

A parallel processing scheme is described for robot controlcomputation on the MIMD shared memory multi-processor model. Since thedynamic control law of a robot arm usually contains a large amount ofoperations compared with the number of available processors, it isimportant to consider not only the effect of parallel processing butalso the efficiency of serial processing in each of the processors. Inorder to obtain such a desirable solution for the complex schedulingproblem, optimization and quasi-optimization algorithms are proposed.The excellent optimization algorithm is based on a branch-and-boundmethod. On the other hand, the practical quasi-optimization algorithm isbased on a partial enumeration method which effectively combines theoptimization algorithm and the approximation algorithm (greatest commonfactor/longest calculation time (GCF/LPT)) that was developed by theauthors