Foundational Simulation Modules for Future CMSA Safety Benefit Assessments

This work presents six foundational fast-time simulation modules for the future safety benefit study of UAS Traffic Management (UTM) Services, such as Conformance Monitoring for Situational Awareness (CMSA). First, a methodology is proposed to translate simulated or real data into Operational Volume Block (OVB) dimensions that comply with the ASTM conformance requirement of 95%. Fixing OVB blocks in length or duration led to the emergence of temporal and spatial overlap which are associated with contiguous and overlapping OVBs, respectively.

Second, realistic scenarios are automatically created by randomly and periodically selecting geographic locations (e.g., San Francisco or Houston) and spawning services that follow operational profiles that have been linked to actual geographic features. Here, we implemented hub-and-spoke, out-and-back and point-to-point operational profiles that can be connected to package delivery, inspection, and emergency medical supply transport services, respectively. Those services translate into flight paths that populate the airspace. Consequently, now practically an infinite number of scenarios can be simulated and it is possible to run Monte Carlo simulations whose conclusions can be generalized to the complete National Airspace System and will not potentially overfit to a few manually designed airspace geometries.

Third, it is shown how the dimensions of the OVB sizing stage are used to discretize the flight paths of the scenario generation, even around waypoints where the direction of flight changes.

Fourth, a low-latency 4D volume-based conflict detection module is proposed that uses R-trees. Three different implementations are considered. Results show that R-trees created with the OVBs of all flights in the airspace consistently lead to the lowest computational time when detecting conflicts, both during the initial airspace configuration and when incrementally adding new flights.

Fifth, a modified Conflict Based Search (CBS) approach is suggested for pre-departure 4D volume-based strategic temporal deconfliction. This method leverages R-tree-based conflict detection at the low level and employs a Constraint Tree structure at the high level. The cost function integrates both total airspace delay and the number of conflicts, with a higher weighting placed on conflict reduction, as achieving a conflict-free solution is the primary objective. Results show that it is able to resolve all conflicts in randomly generated scenarios, even when flight trajectories are diverse and unpredictable, including crossing, merging, and parallel flight paths. Moreover, it successfully deconflicts complex flight plans introduced into an already conflict-free airspace within seconds. Parallelizing the CBS algorithm introduces a beam search-like effect, which contributes to lower overall airspace delays by exploring multiple solution paths concurrently. However, this benefit comes with a trade-off: the computational cost increases compared to the standard sequential greedy strategy.

Sixth, a scenario exporter module that allows researchers to import deconflicted scenarios in JSON format for use in any fast-time simulator that supports the ASTM UTM Protocol for OVB definition. The ultimate goal is to enable any simulator to import realistic, high-fidelity scenarios, allowing researchers to focus on the development and analysis of current and more advanced UTM functions, such as CMSA, instead of repeatedly building some of these supporting tools from scratch.