Analysing the effect of uncertainty in airport surface operations.

Designing efficient and robust automated planning and scheduling systems for complex logistics applications presents many challenges, especially if the executing environment is changing unpredictably. In this paper, we study the problem of planning and scheduling airport surface movement at large airports. We describe a fast-time simulation tool that is easily extensible to different scenarios and airports, and is capable of simulating uncertainty in the form of unexpected delays in aircraft movement at different locations. We also describe an approach to scheduling surface movement which is comprised of a deterministic 'rolling horizon' scheduler, paired with heuristic search based on conflict detection and resolution. We summarize a set of experiments designed to quantify the effects of uncertainty on the robustness of schedules and the incurred delays. The system developed is open sourced at https://github.com/heronyang/airport-simulation.