ACSS Among Teams Advancing Collision Avoidance for Unmanned Aircraft
The avionics manufacturer has joined General Atomics and Northrop Grumman to test ACAS Xu software
Northrop Grumman’s optionally piloted Firebird aircraft participated in flight tests conducted last August in conjunction with ACSS, which manufactures the T3CAS traffic management computer. The tests were part of the Run 2 phase at Mojave, California.

Testers from industry, government and academia planned to resume flight tests this summer of a detect-and-avoid capability for large unmanned aircraft systems (UAS) using next-generation collision avoidance technology. The tests are contributing to the goal of introducing UAS into civilian, unrestricted airspace and development of the Aircraft Collision Avoidance System for NextGen (ACAS X)—technology the U.S. Federal Aviation Administration plans as a replacement for the current TCAS system required on large passenger-carrying aircraft.

Avionics manufacturer ACSS, the joint venture company of L3 Communications (Chalet 306, Static Display C2) and Thales (Chalet 263, Static Display B1) based in Phoenix, supplied its T3CAS traffic management computer to host developmental ACAS Xu software for separate flight tests last year with UAS manufacturers Northrop Grumman and General Atomics. Designed specifically for unmanned aircraft, ACAS Xu is one in a family of ACAS systems developed by the Massachusetts Institute of Technology Lincoln Laboratory and Johns Hopkins University Applied Physics Laboratory.

Participating companies planned to conduct a “Run 3” series of flight tests from June to August at Edwards Air Force base in California. “We’re incrementally testing the different logic upgrades of ACAS Xu,” explained Greg Boerwinkle, ACSS’s UAS program manager. “After each round of testing we collect data and that helps the FAA and Johns Hopkins and MIT Lincoln Labs to improve the algorithm.” ACSS has used internal research and development money to fund its involvement.

During the Run 2 phase, ACSS participated in collision avoidance flight tests with Northrop Grumman last August at Mojave Air and Space Port, California, making use of the latter company’s Firebird optionally piloted aircraft. It participated with General Atomics at Gray Butte Field, California, for flight tests of a Predator B UAS that concluded in November.

The testers flew various encounter scenarios pitting the Predator B and the Firebird with a safety pilot aboard against manned “intruder” aircraft to evaluate the performance of ACAS Xu to detect potential conflicts and recommend maneuvers to prevent a collision. ACSS flew a King Air intruder aircraft fitted with an avionics rack in the cabin to validate that ACAS Xu processed inputs from participating transponders; it also was able to calibrate the system’s performance with GPS position truth data.

ACAS X uses transponder interrogations as well as automatic dependent surveillance-broadcast (ADS-B) signals to detect and track nearby aircraft; it is also compatible with electro-optical and infrared sensors. Whereas current TCAS II technology uses rule-based logic, “ACAS X’s logic employs probabilistic models to represent various sources of uncertainty,” according to MIT Lincoln Laboratory. “To compensate for imperfect sensors, a surveillance and tracking module explicitly takes measurement and dynamic uncertainty into account by representing relative positions and velocities as a probabilistic state distribution. To assess potential collision risks, ACAS X uses computer-optimized logic lookup tables that capture each possible state in the probabilistic state distribution.”

Explained Boerwinkle: “They’re essentially a seven dimensional look-up table based on the number of parameters—closure speed, size, altitude—and all of these look-up tables require a larger memory. The ability to access any point in that seven-dimensional table at any time is really requiring us to update our hardware. It’s much different than the way TCAS is implemented, and our hardware is designed around TCAS. It’s a changing of the way computations are done.”

ACAS Xu for unmanned aircraft will require a database potentially as high as 16 gigabytes, orders of magnitude more than the memory on current TCAS units, which are capable of ADS-B functionality. “With that amount of data, there has to be a validation criteria also,” said ACSS president Terry Flaishans. “Validating that amount, and the number of combinations and permutations is quite large. What Johns Hopkins is trying to do is use simulation to show all the corner cases to make sure that all the different cases are taken care of. The other thing ACAS does is it allows you to do lateral maneuvers along with vertical; TCAS right now only does vertical.”

The current T3CAS unit has enough processing throughput for both Run 2 and Run 3 testing, but would have to be modified for full ACAS capability, Flaishans said. “The memory size itself is not an issue—it’s the modification, qualification and testing of the scenarios that require validation,” he added.

This and other work, including detect-and-avoid flight tests that NASA, the FAA, General Atomics, Honeywell and BAE Systems have conducted using NASA’s Ikhana Predator B testbed, are contributing to standards being developed by FAA advisory organization RTCA. Two RTCA special committees—SC-228 for UAS minimum operational performance standards; and SC-147 for TCAS—are collaborating on the development of detect-and-avoid capability for unmanned aircraft. The goal is to publish a standard by 2020 that calls out ACAS Xu for collision avoidance.

ACSS executives were encouraged by the results they’ve seen from the testing thus far. Last year, the team planned to use the intruder flights only for preliminary functional testing, with the Run 2 phase scheduled for this spring. “But because last year’s testing went so well and we gathered such good data from it, we were able to accelerate the test schedule to do Run 3 this year,” Boerwinkle said. “We advanced our test schedule by about 10 months.”

While some corner-case errors were experienced in the detect-and-avoid system’s performance, Flaishans said the test results he had seen exceeded his expectations.

“If we can get this into a manageable situation where the safety is just as good as see-and-avoid, they could be using these airplanes to do border patrol, search and rescue, powerline (inspection) and not have to have a pilot in the airplane,” Flaishans said. “Over time, UASs will be integrated into the airspace and I believe we’ll be a good part of that.”