Aviation Week was recently invited to fly on a test flight of the U.S. Air Force’s Auto-ICAS (Automatic Integrated Collision Avoidance System), the world’s first combined air and ground collision avoidance system for combat aircraft.

The Auto-ICAS builds on earlier development of the Auto-GCAS (Automatic Ground Collision Avoidance System) which Aviation Week last flight-tested in 2010, and the Auto-ACAS (Automatic Air Collision Avoidance System) that we evaluated in 2014. The combined systems form Auto-ICAS, which has been in testing at Edwards AFB in California since 2016, following an initial evaluation by the U.S. Air Force Test Pilots School in 2015.

Since its introduction three years ago, Auto-GCAS has already saved seven pilots and six F-16s (including a two-seat F-16D), and developers hope the follow-on Auto-ICAS will provide even more protection. The Air Force Research Laboratory (AFRL) together with its collision-avoidance development partners at the Office of the Secretary of Defense, NASA, Air Combat Command, Air Force Test Center and Lockheed Martin, predict substantial savings in both pilots and aircraft if the integrated system is installed in the F-16 and F-35 fleets. Through 2040, the analysis forecasts that 40 pilots and 57 aircraft worth $6.73 billion will be saved.

My aircraft, "Skull 1," was an F-16 Block 50 of the 416th Flight Test Sqdn. flown by Maj. Jameel Janjua, a Royal Canadian Air Force exchange officer and lead test pilot for the Automatic Collision Avoidance Technologies (ACAT) Fighter Risk Reduction Program. As the "shooter" aircraft, we would play the more dynamic role in interchanges with a "target" F-16, Skull 2, flown by Air Force test pilot Brian Kelly with AFRL ACAT expert Lt. Col. (ret.) Kevin Price as observer.

With our altitudes verified, we began a series of automatic collision avoidance maneuvers with very little altitude separation. For the first run at 20,000 ft. we began 4,000 ft. behind and 200 ft. below Skull 2, which was flying at 300 kt. We accelerated to 390 kt. and closed in while maintaining one wingspan separation. Believing we were at the same level as Skull 2, the Auto-ACAS activated, and we performed a bunt while the other F-16 pulled up. “We are trying to run one aircraft into another in the virtual world, and for whatever reason—and we are talking miniscule differences in altitude—the system thought it made more sense for this guy to go up and for us to go down,” says Janjua.

Taken on a previous test flight, this clip (above) shows the collision-avoidance system activating after the "shooter" approaches too quickly and too close from directly astern. The clip includes the view from a camera worn by the pilot as well as footage from the head-up display and specially created 3D animation to provide a virtual picture of what the system thinks is really happening.

The test provides system designers with quantitative data on whether activation is at the appropriate time, if it should have activated at all, or whether the reaction was sufficiently aggressive or too early. Either way, despite the altitude offset, Skull 2 seemed to loom alarmingly close as we closed in at 90 kt. “It is very eye-opening,” agrees Janjua. “Even though you might be thinking, ‘Holy cow! That was close," you are still 200 ft. farther away than you should be. As a fighter pilot who has been flying for a long time, you don’t usually see anyone else that close to you.”

The next tests (similar to this previously recorded test above on an earlier mission) involved closing in on the target at 425 kt. and 20,000 ft. In our case, we banked 15 deg to the left. This resulted in an activation for us to the left and for the target to the right. We followed this by executing a second low-angle crossing encounter in which Skull 2 closed in on our aircraft, flying 200 ft. above us and decelerating as it banked 20 deg. to the left and crossed our virtual bow. On the first attempt, both aircraft violated the exclusion zones so, although the system activated, neither of us maneuvered. Two repeats of this run resulted in more activations, but only Skull 2’s F-16 was triggered into a roll and pull to the right as, on both occasions, we violated the exclusion zone.

Next we descended below 5,000 ft. for the "graduation exercise," the staged head-on collision scenario in the saddle between the two mountains, Garlock Peak to our left and Skulls Peak to the right. For safety reasons, being so close to the terrain, the altitude split between our aircraft was increased to 1,200 ft., even though in the virtual world the system thought we were at the same altitude.

In this video clip (above), both aircraft bank to the right and pull 5g in a climbing turn; note the passage of the opposing F-16 at 24 sec. Skulls Peak loomed suddenly to our right, but the combined anti-collision system was instantly aware of the ground threat and kept us at a safe distance. “We saw a lot of ground as it skirted the peak and flew between the mountain and the other aircraft. That’s essentially what we are trying to do—we are trying to force the system to choose between the airplane and the ground, and the smart decision is that it will avoid both,” says Janjua.

In this clip, taken on another test flight involving four aircraft, we see the combined response as all Auto-ICAS systems activate in close proximity to the mountains on either side.

For our next test point, we again approached the saddle, this time to evaluate the response of the system while conducting a close proximity overtake within the narrow confines of the valley. Beginning the run 4,500 ft. astern and 1,200 ft. below Skull 2, we accelerated to 390 kt. to prompt a system activation as we passed between the two mountains. This time, again with the system thinking we were at the same level in the virtual world, it commanded our aircraft to fly up.  A similar scenario is illustrated in this video clip:

To extend our test flight, we rendezvoused with Ghost 66, a KC-135R, to each take on 3,500 lb. of extra fuel. This was my first experience of air-to-air refueling in an F-16, and I was impressed by the precise flying by both the pilots and the boom operator as the boom nozzle was guided just inches from the cockpit glazing into the receptacle on the upper midfuselage.

Switching gears, we next examined the system’s susceptibility to nuisance activations during operationally representative basic fighter maneuvers (BFM) and normal rejoins, even if the pilots are over-aggressive or making errors. With the collision-avoidance system disabled but still sufficiently active to provide an audible warning in our headsets if an activation would have occurred, we began at 5,000 ft. with a series of gun attacks.

The third BFM maneuver was the most aggressive of all and involved us maintaining a low-aspect close-pursuit guns track. As we got to within 1,300 ft. of the other F-16 Janjua told Skull 2 to execute a 360 deg tuck-under-jink maneuver. Pressing home the attack we corkscrewed and pulled hard as we deliberately got uncomfortably close to his tail. “I purposely pursed him through that jink, and by continuing to pursue him, I’m eating into the range for weapons and safe separation. It’s another area where we have had mishaps with pilots who tracked too aggressively and ended up with a midair collision,” says Janjua.

Flying at 500 mph, we came within 375 ft., which is well inside the 500-ft. training bubble limit and “ever so slightly closer than I’d have wanted to,” Janjua adds. During our corkscrewing pursuit, he had called out the closing range—“900 ft. . . . 700 ft.”—but a fraction of a second later we were very much closer.  “We went through his wake, so I had to be careful of jetwash, and I eased off the stick as we went through the wake from his wings and jet because I did not want to overstress the aircraft,” Janjua says.

From a test-pilot perspective, this is a particularly unnatural part of the evaluation. “You are forcing yourself to make gross errors to the point where you can prod this bear but at the same time stay far enough away to avoid a collision,” says Janjua. From a nuisance-evaluation perspective, the exercise was a success. Throughout the gun tracking tasks, there was not a single tone in our headsets, indicating the system never wanted to activate.

Testing gunfight scenarios without nuisance activations is pivotal to the acceptance of a system that will not impede operations. “I understand because I’m a fighter pilot who has flown in combat. That’s why this side of the testing is just as important because as soon as we have someone who makes a conscious decision to turn off the system, then we have failed. We cannot field a system that’s prone to nuisance because if someone turns it off, what good is it?” says Janjua. 

“We make sure we explore these dark corners because people are more likely to trust it as a credible system and use it long enough to realize it is going to save them and the lives of their friends. Like Auto-GCAS is doing, as soon as you start to save lives, the system gains capital,” he adds.

The remainder of the flight included a supersonic head-on collision-avoidance scenario. Just setting up for this test point required a high degree of precision flying from both test pilots as our aircraft climbed to 30,000 ft. and 31,000 ft., respectively, before beginning to run in supersonically toward each other from a distance of about 40 mi. Accelerating to Mach 1.2, we covered the ground rapidly with a closure speed of Mach 2.4, or around 24 mi. per minute.

Everything seemed to happen very quickly as I watched the fuel flow gauge numbers tumble with a blur and the Mach meter indicate our supersonic velocity of 1,200 ft. per second.  Trying to pass each other virtually within a wingspan, I watched as the system’s warning chevrons appeared and touched, instantly rolling us to the right and then pulling us into a 5g turn. With an altitude offset of 950 ft. for safety programmed into our Auto-ACAS system, I did not expect to even see the other F-16 but was surprised to glimpse a gray streak as Skull 2 passed on our left and activated in concert. Post-flight analysis later indicated we had passed virtually with a missed distance of 352 ft. 

Starting with a low-level run to simulate tactical terrain masking and threat avoidance, we dipped down to 300 ft., our shadow looming large as we swooped at 450 kt. over the sage bushes dotting the valley floor. Although an “altitude, altitude” warning blurted out as we clipped a ridge at 250 ft., no chevrons appeared on the HUD. “It just shows how robust the Auto-GCAS is, even though we flew at such high speed and low altitude over ridges and terrain that was flat or sloping terrain. This will wait until the last infinitesimal second before the point of no return,” says Janjua. 

Although this video was taken during my initial Auto-GCAS test flight it accurately represents a similar phase of low-level flying conducted during my ICAS sortie to stress the combined system.

We then returned to Edwards, landing after a 2.7-hr. sortie which proved to me that the Auto-ICAS system developed and tested here has not only dramatically expanded the potential flight safety of combat pilots but has also in no way changed the functionality of the baseline Auto-GCAS on the F-16 Block 40/50.