Agent-Based Modeling of Uncrewed Aircraft System Flight Planning for Airspace Fairness

The number of uncrewed aircraft systems (UAS) and corresponding UAS operations is expected to increase dramatically soon. Regulatory agencies are currently designing and piloting UAS traffic management concepts that rely on federated protocols and a cooperative, community-based approach. As demand increases, the need to ensure fair usage of airspace among operators will be an important challenge. Currently, there are no widely agreed upon definitions or guidelines for UAS airspace fairness. The objective of this work was to evaluate the fairness implications of using a first-filed-first-served protocol for flight planning. We developed UAS Cooperative Airspace Traffic Simulation (UCATS™), an agent-based modeling simulation tool, to evaluate different UAS package delivery scenarios using a first-filed-first-served approach to UAS flight planning. We defined key metrics to measure fairness, including average delay, maximum delay, and percentages of flights as planned, replanned, and canceled. Our results showed that a first-filed-first-served approach may cause flights having a departure time later in the day and flights that are filed with less advanced time have a higher probability of experiencing more negative flight outcomes. We also evaluated the sensitivity of fairness metrics to different traffic levels, different flight densities, and the addition of food delivery operations.

Introduction

As demand for uncrewed aircraft systems (UAS) operations increases, regulatory agencies must develop concepts for managing widespread UAS operations. In the US, future UAS traffic management (UTM) will be community based and cooperative. This approach differs from traditional crewed air traffic management, which relies on the Federal Aviation Administration (FAA) to provide centralized traffic management. In UTM, the FAA will establish “rules of the road,” but the operators and thirdparty UAS service providers will be responsible for coordination, execution, and management of operations. 

With the decentralization of UTM, competing operators should work together to ensure that all have fair access to the airspace. The FAA will need to ensure that UTM is implemented fairly, despite the competing interests and unequal market shares among operators. Currently, however, the FAA has yet to formally define what fair access to airspace entails.

In this work, CNA developed UAS Cooperative Airspace Traffic Simulation (UCATS™) as a tool to investigate UAS airspace fairness with the intent of providing insight into future decision-making for UTM. UCATS™ is an agentbased modeling (ABM) tool that simulates UAS flight planning scenarios to provide insights into usage of airspace, including metrics to measure fairness. Industry and government stakeholders can use this tool to assess future high-density UAS operations and resulting fairness toward UAS operators. 

PRIOR WORK

There are limited studies of airspace fairness for UAS operations. Evans et al. (2020) simulated UAS scenarios where two operators served in overlapping regions. The study used a first-come-first-served allocation of resources with de-confliction by departure time and determined an optimized solution set based on the cost of delay to the operator. The study found that significant imbalances in delays occurred due to shorter file-ahead times and higher traffic levels.

Sacharny et al. (2020) compared the efficiency of two UTM strategic deconfliction scenarios: a gridded approach, where the airspace is defined into grids that each must be deconflicted independently, versus a lane-based approach where deconfliction only occurs on predefined lanes of traffic. The study used a series of computational experiments to determine relative metrics of average and maximum delay, flight time, and computational deconfliction time to determine that the lane-based approach outperformed the gridded approach.

Finally, Chin et al. conducted a series of studies on efficiency and fairness for future UAS operations. The first study simulated the problem of four package delivery operators making overlapping deliveries with different fairness metrics. The study found that total delay was a more effective metric than departure reversals. The second study expanded this problem to consider operator preferences, airborne-to-ground delay cost ratios, and market shares and found similar results as the previous study. A third study computed optimized solution sets based on queuing and lane-based fairness protocols for air taxis and ranked each by the resultant system delays. 

Each of these studies has advanced the field of UAS airspace fairness; however, our study fills technical gaps in the body of work by considering a non-homogenous set of operators that differ in file-ahead time and peak time of delivery, including the reality of flight cancelations, and conducting sensitivity analyses of different traffic levels and allowable flight densities. We were able to incorporate these new considerations into an UAS simulation because of our use of ABM and the flxibility that such a modeling approach provides.

ABM APPROACH

ABM is a bottom-up computational approach used to analyze the effects of autonomous interacting agents on the overall system. Agents are given predetermined properties and interact with each other and their environments using predefined rules. ABM can simulate heterogeneous systems where the behavior of each agent contributes to the overall outcome. In our approach, the UAS operators are the agents, who are autonomous and interact with each other. The operators also follow defined behaviors regarding the flight planning process, eligible trajectories, and conflict detection and resolution procedures.

ABM has been used to simulate a variety of environments, including those related to UAS. Recent studies have addressed specific problems without an explicit focus on fairness, such as lastmile delivery. In a 2019 literature review of ABM applications for unmanned aircraft vehicles, the authors reviewed 42 papers that addressed a variety of UAS topics, none of which directly address strategic flight planning or airspace fairness. We addressed this gap by using an ABM approach that considers the sensitivity of UAS parameters that contribute to strategic flight planning.

OBJECTIVES

Our goal was to develop an ABM tool that simulates flight planning for small UAS delivery operations and use its derived statistical metrics to evaluate airspace usage fairness. This work focuses on two research questions. 

First, how can airspace fairness be measured for small UAS package delivery operations? Metrics such as the average delay and number of flight cancelations per day are helpful in determining how often operations are being completed as planned. Moreover, a relatively fair airspace would have low percentages of cancelations and delays because operators would complete their flights with minimal conflicts.

Second, how will prioritizing flights using a first-filed-first-served (FFFS) method affect airspace fairness? Within this scope, we aimed to quantify how different levels of traffic and types of UAS operations will affect a fair approach to traffic management planning. Eventually, operations will reach a traffic threshold at which an increase in cancelations and delays occurs. Introducing different types of UAS operations, such as food delivery (which naturally operates on a different demand schedule), influences both traffic level and package delivery filing. These variables would impact our metrics for airspace fairness under the FFFS method.