arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1304.3085cs.AI
收藏

未来事件序列中的计划、调度和不确定性

Planning, Scheduling, and Uncertainty in the Sequence of Future Events

Artificial Intelligence Technology Group(人工智能技术组);FMC Corporation(FMC公司)

B. R. Fox, Karl G. Kempf

摘要:

工厂环境中的调度由于计算复杂性和时间不确定性而变得复杂。总之,这两个因素保证了构建最佳时间表的过程将是昂贵的,并且执行该时间表的机会将是微小的。任务执行时间的时间不确定性可以通过几种方法来抵消:通过仔细的工程来消除不确定性,在确定性丢失时恢复确定性,通过使用更精确的传感器来减少不确定性,以及量化和限制剩余的不确定性。不幸的是,这些方法只集中在不确定性的来源,并未能应用知识的任务,这是调度。一个完整的解决方案必须根据生产世界的发展状态调整要执行的活动的时间表。视觉导向组装的例子来说明,最小承诺的原则,在创建一个计划,在表示的时间表,并在执行的时间表,使机器人智能和有效地操作,即使在存在相当大的不确定性的未来事件的顺序。

英文摘要:

Scheduling in the factory setting is compounded by computational complexity and temporal uncertainty. Together, these two factors guarantee that the process of constructing an optimal schedule will be costly and the chances of executing that schedule will be slight. Temporal uncertainty in the task execution time can be offset by several methods: eliminate uncertainty by careful engineering, restore certainty whenever it is lost, reduce the uncertainty by using more accurate sensors, and quantify and circumscribe the remaining uncertainty. Unfortunately, these methods focus exclusively on the sources of uncertainty and fail to apply knowledge of the tasks which are to be scheduled. A complete solution must adapt the schedule of activities to be performed according to the evolving state of the production world. The example of vision-directed assembly is presented to illustrate that the principle of least commitment, in the creation of a plan, in the representation of a schedule, and in the execution of a schedule, enables a robot to operate intelligently and efficiently, even in the presence of considerable uncertainty in the sequence of future events.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑