| dc.description.abstract | With the rapid growth in the number of uncrewed aircraft systems (UAS) being deployed for on-demand delivery applications, severe congestion is anticipated in the low-altitude class-G urban airspace. Thus, there is a need for structuring the urban airspace and establishing a controlled UAS Traffic Management (UTM) system to mitigate this congestion. In this context, CORRIDRONE, an adaptive multilane UTM architecture and its concept of operations, were proposed in the literature. This thesis addresses the problem of congestion mitigation under time-recurring delivery demand in the presence of static obstacles and dynamic wind in the CORRIDRONE framework.
Delivery applications typically require UAS to travel between source-destination (S-D) pairs on preplanned paths. Geofencing these paths can physically separate a UAS trajectory from other UAS and static obstacles. However, with increasing demand, intersecting UAS flight paths are inevitable, resulting in growing conflicts among the UAS. Conflicts in confined, dense traffic regions would be detrimental to UAS safety and compromise throughput. Such limitations are mainly due to treating UAS traffic as a road-like network or air traffic, where UAS are restricted to preplanned paths, a preplanned number of lanes, and fixed air volumes.
In this thesis, we propose a multi-stage path planner, wherein numerical potential fields are used to periodically replan flight paths as the UAS transits between the desired source and destination. The generated flight paths are sub-optimal and maintain an optimum stand-off distance from the static obstacles and congested regions. For UAS to follow these paths, a dynamic cylindrical geofence volume moving along the flight path is designed, wherein the operational air volume is sized considering the UAS’s compliance and mission-specific requirements, as well as environmental wind conditions. An adaptive artificial potential field encompassing the moving geofence volume is proposed that is responsible for both UAS path following and UAS safe-keeping within the geofence volume. With APF-embedded geofence, the UAS does not interfere with other UAS sharing the airspace. However, to ensure UAS’s safety under threats from dynamic known and unknown intruders in the vicinity, an octant-based local collision avoidance scheme is proposed.
A traffic simulator to study the effectiveness of the proposed path planner has been developed, considering time-recurring delivery demand between a source-destination pair. Time-recurring delivery demand results in continuous UAS traffic flowing on the S-D route’s nominal path. The above path planner effectively steers the UAS from exogenous congestion occurrences on other S-D routes. However, in the event of congestion within the given route, the traffic flow excessively spreads around its nominal path.
To overcome this problem, a heading-constrained airspace design and a distributed rule-based congestion mitigation strategy are proposed in this thesis. Individual UAS decisively utilize the airspace surrounding the nominal path and dynamically generate local paths. These paths are opted for when the nominal paths are congested. The UAS reverts to the nominal paths once the congestion reduces. At the microscopic level, individual UAS mitigate congestion; however, at the macroscopic level, the system-level behaviour resembles a dynamic number of parallel air lanes being activated/deactivated as a function of the traffic inflow rate. Discrete-time Queuing theory is adopted to estimate the expected air volume that would be utilized by UAS as a function of the probabilistic occurrence of congestion when the
demand is stochastic and time-recurring. The UTM can adaptively reserve such air volumes surrounding the nominal path beforehand to accommodate any foreseen congestion.
Finally, we consider time-recurring delivery demand from multiple stakeholders, resulting in non-homogeneous UAS traffic flow on a given S-D route. Multiple feasible paths may exist between the source and destination. However, all stakeholders may simultaneously opt for the optimal path, resulting in congestion on this path and increased travel time. We propose a restless bandit-based congestion pricing and path reservation system to mitigate this congestion. | en_US |
