Conceptual-Level Analysis and Design of Unmanned Air Traffic Management Systems
Ramee, Coline L.
MetadataShow full item record
There have been multiple announcements by different companies in the past couple years of package delivery by drone and air taxi projects. However, there are still many barriers to the deployment of high densities of aerial vehicles in low-altitude airspace over urban areas. Current Air Traffic Control Systems cannot handle the high density of traffic being forecast. Integrating these new types of on-demand air mobility in the National Airspace requires a fundamental change to the traffic management system. Many different concepts of operations for unmanned traffic management (UTM) systems have been proposed, but there is no common framework to evaluate and compare alternatives at a conceptual design stage. This might cause a locally optimal system to be chosen, resulting in lower safety and economic performance than what would have been possible if a more systematic approach to the design of UTM system had been followed. In this thesis, a systematic approach to the design of UTM systems is introduced. Based on the literature on conceptual design, a five step approach to the design of UTM systems is proposed. The steps of the approach are: define operating scenarios, generate UTM alternatives, select performance criteria, evaluate, and make decision. To generate UTM alternatives in a systematic manner, a matrix of alternatives should be created. However, this requires a system decomposition that does not currently exist for UTM systems. Here, a system decomposition into four subsystems is proposed: airspace structure, access control, preflight planning, and collision avoidance. For each subsystem, alternatives are identified using the literature. For the second step of the approach, operating scenarios for UTM are not well-defined. There are many external factors outside of the designer’s control, and different studies make different assumptions. Three different external factors, or components of an operating scenario, are identified: demand, static obstacles, and priority traffic. The impact of the different subsystems and external factors on the performance of a given UTM architecture cannot be found in the literature. Many studies evaluate a point design or fix assumptions to focus on a single subsystem. There is no available tool that allows to evaluate different UTM architectures while varying all the elements that have been presented here. To bridge that gap, an agent-based simulation was developed to allow the evaluation of the UTM systems generated using the matrix of alternatives in different operating scenarios. For the fourth step of the approach, performance criteria are selected from the aviation literature. To capture safety, the number of losses of separation and near-midair-collisions per flight hour are used. To measure the efficiency of the trajectories, a time and energy efficiency metrics are introduced. The capacity of the system is evaluated for a fixed overall density using the throughput, or number of vehicles completing a flight per minute. Finally, two simple multi-attribute decision making methods are selected from the literature. This allows to rank architectures based on their performance in a given scenario for a given set of weights representing a designer’s preferences. This thesis also proposes a novel 4D trajectory planning algorithm that relies on a local collision avoidance method. Experiments show that it performs well in terms of time efficiency and throughput when compared to a decoupled approach. The novel algorithm achieves a comparable performance to a global optimization algorithm in a nominal cruise scenario but is much more computationally efficient. The impact of the inclusion of certain subsystems and external factors on the outcome of the conceptual design stage is systematically evaluated in a series of experiments. Performance of different architectures is evaluated with and without the subsystem or external factor of interest. The experiments show that there are significant interactions between agents' autonomous behaviors, airspace structure, and external factors such as demand, static obstacles, and priority traffic. The decision tables obtained with and without the element of interest are compared, and weights are found such that the architecture rankings are different. This shows that neglecting these interactions or making simplifying assumptions may change the outcome of the conceptual design stage and result in the selection of an architecture that underperforms in terms of safety, capacity or efficiency. This is validated on two use cases, an air taxi scenario and a drone delivery scenario. In the air taxi scenario, using the proposed approach results in the selection of an alternative with a 25\% higher score than the alternative selected with a baseline approach. As a result of the work conducted in this thesis, the importance of including the autonomy, airspace structure, demand, static obstacles, and priority traffic in the early stage of UTM evaluation has been demonstrated. The necessity of including other subsystems or external factors can be evaluated by following the same process that was demonstrated in the thesis.