In International Conference on Automated Planning and Scheduling (ICAPS 2020), Nancy, France, October, 2020. Paper abstract bibtex
Generating and scheduling activities is particularly challenging when considering both consumptive resources and complex resource interactions such as time-dependent resource usage. We present three methods of determining valid temporal placement intervals for an activity in a temporally grounded plan in the presence of such constraints. We introduce the Max Duration and Probe algorithms which are sound, but incomplete, and the Linear algorithm which is sound and complete for linear rate resource consumption. We apply these techniques to the problem of scheduling awake and asleep episodes for a planetary rover where the awake durations are affected by scheduled activities. We demonstrate how the Probe algorithm performs competitively with the Linear algorithm given an advantageous problem space and well-defined heuristics. We show that the Probe and Linear algorithms outperform the Max Duration algorithm empirically. We then present the runtime differences between the three algorithms. The Probe algorithm is currently base-lined for use in the onboard scheduler for NASA’s next planetary rover, the Mars 2020 rover.