摘要： 针对移动机器人导航控制中信息处理量大、任务多的情况,提出了一个适用于移动机器人的分布式计算框架,并在此框架的基础上设计了一种任务调度方法——GMBSA,该方法以资源代理为基础,首先对任务执行时间进行预测,然后运用遗传算法结合多队列Backfilling方法进行任务调度,达到最小化任务执行时间的要求,最终实现资源的优化分配,满足了机器人导航控制中的实时性要求。该文采用实验室构建的分布式计算环境对GMBSA的性能进行了测试,并比较了轻重负载情况下GMBSA,多队列Backfilling和FCFS3种调度方案的性能差异。

关键词: 任务调度, 资源代理, 遗传算法, 多队列Backfilling

Abstract: This paper puts forward a job scheduling method of the distributed computing system of mobile robot for the situation that the large amounts of information process and tasks in the navigation control of mobile robot——GMBSA which is based on resource broker.Firstly,it predicts the run times of job,then uses genetic algorithm combining with multiple-queue Backfilling to schedule jobs,and the method optimizes the resources allocation,satisfies the need of real-time.In the end,the paper evaluates the performance of GMBSA,and compares its performance with multiple-queue Backfilling scheduling algorithm and FCFS scheduling policy.

中图分类号: 

• TP242