Guest Lecture Max Li

Title: Optimization Across Scales for Advanced Air Mobility: From Network Management to Adaptive Trajectories

Abstract: Advanced Air Mobility and new uncrewed aerial systems (UAS) operations will require coordinated decision-making across multiple spatial and temporal scales. At the system level, traffic management strategies must balance uncertain demand against limited airspace and vertiport capacity. At the operator level, drone launch and recovery locations, vehicle assignments, and routes must be selected under tight energy and computational constraints. During execution, individual UAS may also need to revise their trajectories as new information becomes available. Treating these decisions independently can produce plans that are mathematically attractive but computationally impractical or operationally brittle.

This seminar presents three optimization frameworks addressing different layers of this broader challenge. The first uses bi-level optimization to represent interactions among fleet operators, airspace service providers, and system-level governance under time-varying and unscheduled AAM demand. A neural network surrogate enables the resulting model to be solved tractably while supporting the evaluation of alternative congestion management strategies. The second examines the coupling between launch-and-recovery siting and energy-constrained coverage routing for agricultural drone operations. Through facility location heuristics and tiered spatial decomposition, the framework generates operational plans on minute-scale timelines with relatively small losses in coverage. The third integrates mission-level routing with informative B-spline trajectory planning for UAS monitoring under incomplete and evolving hazard information. Streaming measurements update a spatial belief map and trigger online replanning when newly acquired information justifies adapting the remaining trajectory. Together, these studies suggest that scalable decision support for AAM and UAS operations should be hierarchical, uncertainty-aware, and adaptive.