NASA testbed to assess traffic-avoidance tools for unmanned aircraft
's , Mitre and several partners have completed the first in a series of inflight evaluations of cooperative automatic sense-and-avoid (SAA) algorithms for unmanned air systems (UAS), in large part to validate a new testbed that will be used for more advanced trials next year.
Called the limited deployment cooperative airspace project (LD-CAP), the two-year program is designed to give industry and academia developers of SAA algorithms a realistic inflight avenue to evaluate their proposals, which typically rely upon sensor data rather than a pilot's eyes to strategically or tactically separate UAS from other air traffic that may pose collision risks.
Officials say the work is complementary to SAA efforts being conducted by the, U.S. Air Force and others. “What we want to do is create the scientific data the community needs to mitigate [SAA concerns],” says Andy Lacher, UAS integration lead for Mitre. In addition to answering whether an automatic algorithm is feasible, Lacher says the project will help RTCA to develop technical standards for SAA.
SAA is regarded as one of the top priorities—and one of the most difficult issues to solve—in the FAA and's effort to safely integrate manned and unmanned aircraft in civil airspace. Partners in the LD-CAP project, all of whom are using internal funding for the effort, include Mitre, the University of North Dakota (UND), Draper Laboratory and the North Dakota National Guard.
During a two-week deployment in early September,flew a modified Cirrus SR22 as a surrogate UAS, with a safety pilot on board, in 147 convergence scenarios with an unmodified UND-supplied 172 acting as the manned intruder aircraft. Scenarios were tested previously in more than 2 million simulator runs, says NASA.
The aircraft were spaced 2,000 ft. apart vertically for safety, but the Cirrus's Automatic Dependent Surveillance-Broadcast (ADS-B) system was biased by 2,000 ft. to make the avionics believe the aircraft were generally at the same altitude. The two SAA algorithms were allowed to maneuver the aircraft in heading, horizontal speed and vertical speed using an add-on “general purpose” computer in the back of the Cirrus that commands the aircraft's two-axis autopilot and throttle.
Both were equipped with Garmin GDL 90 ADS-B units as the sensors for the algorithms, though the testbed in general is designed to be sensor-agnostic. Two algorithms were tested, one developed by UND and the other by Mitre, and flights were conducted in a region of airspace west of Grand Forks, N.D.
“Part of the test here was not only how well the algorithms work, but us validating our capabilities,” says Frank Jones, the LD-CAP deployment lead for NASA Langley.
Complexity will be increased during flight tests planned for June 2013 in North Dakota, with plans for new algorithms that will handle multiple intruder aircraft and noncooperative targets tracked by primary radar, as well as a flexible task automation “super app” developed by Draper.
The super application is based on Draper's Timeliner engine, originally developed in 1981 and now used, among other tasks, for control and sequencing of International Space Station payload bay experiments and control moment gyros. When installed on the Cirrus, the super app will allow multiple algorithms from multiple providers to run in parallel, with Timeliner making command-level decisions based on user-defined parameters. For example, a variety of SAA apps can be running simultaneously with terrain and weather avoidance apps to determine the safest vehicle path with many constraints.
Draper plans to test the algorithms on Mitre's “fast-time” simulator later this year, followed by integration onto the SR22 for the demonstrations in North Dakota in June.