The Auto-GCAS’s capabilities can best be appreciated firsthand.

During a 1-hr. 20-min. flight on the Air Force Research Laboratory/NASA F-16 testbed, this Aviation Week editor observed a variety of scenarios, many flown at extremely low altitudes and high speed, designed to show how the automatic ground collision avoidance system (Auto-GCAS) will prevent mishaps while allowing the full range of tactical flying without triggering nuisance fly-ups.

Project test pilot Kevin “Budman” Prosser began the flight with a full-afterburner takeoff from Edwards AFB’s Runway 22R, followed by a vertical climb to 15,000 ft. While providing a spectacular start, the spirited departure also demonstrated the system’s ability to transition seamlessly from standby to active mode during an aggressive takeoff and climb. The system activates when the gear is raised, altitude exceeds 400 ft. and calibrated airspeed (KCAS)exceeds 200 kt. A smooth transition to active is good, because “with an afterburner takeoff you are raising the gear within the noise of the digital terrain elevation database system, and you don’t want to trigger a fly-up at takeoff,” explains Prosser.

The existing PGCAS (predictive ground collision avoidance system) warning device was then set to trigger at 50-ft. AGL. “We can’t set it any lower, but that’s what it needs to be at to provide the least amount of nuisance alerts and the least amount of warning time. We’re trying to project what the operational guy is experiencing, and by about the third “pull-up” warning he’s started to ignore PGCAS. We found it is only valuable in a nuisance-free environment,” he says.

Before the demonstration could start, the basic operation of the system was validated, including the ability to turn it off. Flying straight and level at 320 KCAS and 9,000 ft. MSL, Prosser pushed the F-16 into a 15-deg. dive and 90-deg. bank before manually engaging the system using the gun-trigger. The pilot-activated recovery system (PARS), triggered by a panel-mounted switch in the production-standard Auto-GCAS, will enable pilots to recover in the event of disorientation.

In response, the aircraft immediately entered an automatic fly-up recovery by rolling wings-level and initiating the standard 5g+ pull-up. Prosser then canceled the maneuver by toggling a paddle switch on the sidestick controller. He reestablished the setup conditions and again placed the F-16 in a dive with bank angle so that I, from the backseat position, could first activate PARS and then cancel it.

From a test perspective, it is an important safety step to check the ability of the backseat position to terminate the system; but it is also intended as a key element of the baseline production configuration in both the single- and two-seat fighter. “As part of the ‘do no harm’ doctrine, we have to make sure that if the system screws up, the pilot can turn it off,” says Prosser. “There was a lot of heated debate over how much control the pilot should have,” he adds.

With Auto-GCAS healthy, we proceeded to the first specific system demonstration. Setting up at 16,000 ft. MSL and 460 KTAS (knots true airspeed), Prosser rolled the aircraft inverted and let the nose fall through at 1g into a 15-deg. dive. With the stick released at this point, the aircraft remained at 1g automatically. The idea is to replicate typical conditions after a pilot has experienced g-loc (g-induced loss of consciousness), or disorientation, and is unaware of the aircraft’s downward trajectory.

Prosser then engaged PARS to initiate the fly-up, which immediately rolled us wings level and headed us back uphill with a 5g+ pull-up. “Here we’re looking at the autopilot. Is it ‘over g-ing’ the aircraft, or maybe rolling it too slow? The whole time, the trajectory prediction algorithm is predicting what the recovery will look like, and we’re comparing our actual recovery profile to what that prediction was like,” he says.

The team has spent substantial time perfecting the recovery characteristics. “There are a lot of variables at play—initial bank angle, atmospherics, stores configuration and so on. The F-22 and F-35 will be relatively easy by comparison, but the F-16 has a vast variety of stores loadings.”

Climbing back up to 25,000 ft. and steadying at 310 KCAS, Prosser set up for the next test, which aimed to show how Auto-GCAS would save the day in the event of a miscalculated recovery from a split-S or air combat maneuver. When the aircraft is rolled inverted and the nose pushed down into an almost vertical dive, the autopilot’s predictive algorithms must quickly evaluate whether to pull through or recover in the opposite direction. “The bank angle is almost indeterminate, so you can’t roll wings-level when you’re pointing straight downhill. It becomes a mathematical problem to solve, so we don’t get a ‘divide by zero’ situation with the autopilot,” Prosser says. Having initiated PARS at 4g and 85-deg. pitch attitude, the autopilot did the math successfully and commanded a pull-through.

Prosser took back control and climbed to prepare for the next condition at 12,000 ft. and 440 KTAS. Having evaluated the autopilot’s pull-through logic, this time we looked at the roll-through logic. The F-16 was in a steady 25-deg. dive before Prosser entered full left stick and initiated PARS. Rolling through from a 90-deg. bank position, the system suddenly initiated a roll reversal, having decided “it was faster to recover that way,” says Prosser. On a second dive, the recovery was initiated slightly later in the roll but still prior to reaching inverted, and the system completed the roll-through before initiating the 5g+ pull-up.

Now it was time to wring out the system’s low-level capabilities starting with a simulated strafe run at Edwards AFB’s West Range. Pulling high-g to stay within the tight confines of the range approach, Prosser set the F-16 up for a low-angle strafe at 480 KCAS. “Strafing is one of the two most challenging scenarios for Auto-GCAS. “Strafe is an operational maneuver that gets you the closest to the ground by far—but in this case, over smooth terrain. The second most challenging is low-level masking in rough terrain,” he says.

The main challenge for the system is to safely enable the aircraft to reach extremely low altitudes of 50-100 ft. without triggering nuisance fly-ups. “The logic here is that it’s a good system if it keeps you from hitting the ground and allows you to go as legally low as you could on the range,” Prosser explains. “If it can get you down to 75 ft., then it’s clearly adequate. If you stay in the 50-100-ft. range, it’s still pretty darn low.”

In all, we conducted three runs; an initial strafing down to 140-ft. AGL and two that came down to 120 ft. None triggered a fly-up, despite the eye-watering proximity of the ground and the high speed of the runs.

Leaving the range, we transitioned to a relatively featureless and flat part of the practice area called Cords Road, where Prosser descended to around 300 ft. initially to show the robustness of the system against nuisance fly-ups in “high-protection” mode. This provides more of a built-in altitude buffer to the prediction algorithm for protection mainly against tall trees and, in operational form, will be renamed the “normal” mode. “For operational guys flying at 500 ft., this setting will be good enough almost anywhere, but it won’t work for strafe or low-level over rough terrain,” says Prosser.

To handle these situations, the system has a “low protection” mode, which was then demonstrated during reruns over the barren, sage-dotted Cords Road. Runs were made down to 100 ft. After a straight and level pass, Prosser then deliberately stressed the system by flying back down the area and maneuvering aggressively with a series of S-turns at speeds up to 560 kt. “One of the things we found was difficult with the F-16 was predicting what the roll rates are going to be because the inertia and rates vary from configuration to configuration,” he says.

However, no matter how wild the low-level, high-speed maneuvers seemed to be from the aft cockpit, the system coped with the varying roll rates, and no nuisance fly-ups were triggered. “We are trying to validate that the system is not too conservative on roll,” he adds.

We then transited supersonically to the starting point for the most challenging part of the day’s flight. We were to make a prolonged terrain-masking run through a low-level route from the barren wastes of the Panamint Valley to the snowy tops of the Sierra Nevada. The route encompasses very rugged terrain and provides a major test of the system with precipitous peaks, knife-edge ridges and narrow, winding ravines.

Initially flown in high-protection mode, the F-16 was pushed into the only “nuisance” fly-up of the entire flight as we hurtled up a ridge at the top of Rainbow Canyon at the northwest end of the Paramint Valley. The maneuver momentarily sent us up several hundred feet before Prosser banked the F-16 inverted and, in low-protection mode, dropped down to a significantly lower height for the run across the Owen Valley and up into the high forests of the Sierra Nevada. Software fine-tuning is being finalized to smooth out the type of rare nuisance fly-ups that have been triggered by niche terrain effects such as Rainbow Canyon, says the team.

The true robustness of the system was then demonstrated during a dramatic, gut-wrenching descent down the narrow, twisting Kern River Valley. With ridge and mountain tops on either side, and the river flashing below, Prosser flew the F-16 within the tight confines of the valley at less than 200 ft. and more than 500 kt. without provoking a single fly-up despite the fact the collision avoidance demonstration was flown with elevated “buffers.”

A final set of demonstrations began with a high-angle strafe to replicate a mishap that occurred at Balad, Iraq, in which an F-16 failed to clear terrain beyond the target area. Diving at 27 deg. and 15 deg. of bank, our aircraft was at around 410 KTAS when the fly-up was automatically commanded. Using the digital terrain elevation database, the system calculates the correct recovery profile, taking into account obstacle clearance requirements and aircraft performance.

“It will clear the obstacle with the current power setting, even though the F-16 doesn’t have an auto-throttle. However, if the pilot does the right thing [and increases] the power, it will result in better protection,” says Prosser.

The next demonstration replicated an “air score” mishap in which a pilot on a bombing run crashed after losing situational awareness. Having passed over the target and pulled the nose up, the pilot rolled the aircraft to check the impact of the weapons. However, unknown to him, “while looking over his shoulder, he was really in a bank dive,” says Prosser. To emulate the situation, we entered the run at 11,600 ft. and 440 KTAS, before diving at around 20 deg. with a 150-deg. bank angle. Projecting an imminent ground collision, and sensing no pilot reaction, the system abruptly unloaded the roll to wings level and initiated a 5.5g fly-up.

The following test consisted of low- and high-speed runs against the 1,200-ft.-high Fremont Peak near Edwards dubbed GCAS Mountain by the test team. A 1,300-ft. buffer was built into the test, giving Auto-GCAS the illusion we were flying against a larger mountain.

Having experienced test flights of commercial terrain-awareness and warning systems, the sensation of deliberately aiming at a mountain was not new to me. What was new, however, was the 540-kt. speed at which we closed on the high terrain.

“Commercial systems have a lot of pilot reaction time built in,” explains Prosser. “However, we’re in the 1-1.5-sec. timeframe to fly-up, not the sort of 20-sec. warning time of those systems. So we’re getting an actual fly-up well past where a system would be telling you to ‘pull up, pull up’!” he adds. The time-dependent prediction algorithm determines when the fly-up occurs and what climb profile is needed to clear the obstacle. At lower speed, the fly-up is initiated later than at high speed.

My flight concluded with a demonstration of a low-speed-protection feature developed specially for air show maneuvers. At 10,450 ft. and 195 KCAS, Prosser rolled the F-16 and pulled into a split-S at maximum aft stick.

“Although we do a 2g pull, it’s not enough to complete the maneuver, which is also started slower [around 5 kt. slower than the usual Auto-GCAS minimum operating speed of 200 kt.] than we would want to,” he says.

As this condition is similar to a known air show mishap, the “show” Auto-GCAS mode was developed, but it kicks in only if it senses the pilot is not entering the maneuver correctly. In our case, the system reacted, commanding a reversal fly-up after the F-16 had lost less than 600 ft. of altitude.

With fuel running low, we returned to base. The demonstration thoroughly convinced me that Auto-GCAS will save lives and aircraft without getting in anyone’s way.