REACT-AAM seeks to improve standard collision avoidance approaches to accommodate the higher density utilization of mixed-use airspace envisioned in the Advanced Air Mobility concept. To accomplish this, SSCI proposes to develop mission-aware rerouting so that the flight path is optimized for the mission and multiple intruders.
This Detect and Avoid enhancement adds to a host of prior work in intruder detection, sensor fusion, maneuvering threat prediction, and airborne collision avoidance technologies developed over the past decade at SSCI. In prior work, we built a process for evaluating the airspace utilization models where a specific UAS would be flown, and demonstrated that selecting maneuvers from among a fixed set of motion primitives is sufficient to reach the target level of safety. In those projects, we resolved the highest risk intruder at any given moment. For AAM, we propose to fuse risk metrics from multiple threats as part of the cost function, and consider all simultaneous threats in the avoidance strategy.
We have showed that by adjusting the duration and intensity of the single maneuver that the mission path could be optimized for the most urgent threat. In this work, we seek to extend the guarantees of safety to a set of multi-sequence avoidance maneuvers that are constructed using machine intelligence through genetic algorithms. We have demonstrated as part of the DARPA CODE program that our trajectory solver implementation using evolutionary algorithms is near-instantaneous for real-time objective-based path replanning, and expect that it will be equally adept in the DAA space. Thus, we can improve performance of the mission and meet the metrics for flight safety amid multiple intruders through a heterogeneous series of dynamic responses.
REACT-AAM is applicable to NASA’s UAS integration in the NAS programs, as each UAS will be required to have onboard SAA. The specific challenge will enable Urban Air Mobility and other Advanced Air Mobility components to safely maneuver in congested airspace and resolve real-time conflicts in flight paths. The technology is also are applicable to other conflict avoidance needs, such as required for orbital adjustments to avoid space debris.
REACT-AAM plans amid multiple risk sources and constraints. The approach within REACT-AAM will have applicability to autonomous vehicles that are constrained by traffic laws, have comfort objectives, and must navigate among other vehicles. It could help solve autonomous navigation of narrow shipping channels in the USV domain.