A new light for safety

Despite the brushes with aircraft collisions at U.S. airports, FAA has no plans to expand its most elaborate and expensive ground collision alert technology beyond the 20 airports that already have it. Keith Button spoke to FAA-funded researchers who believe they might have a lower-cost solution.

If all goes as planned, in 2025, San Antonio International Airport in Texas will be the site of a safety experiment that will ultimately involve the flying public, air traffic controllers and pilots. Between now and then, a line of five holes will be cut into some of the taxiways close to where they join the airport’s runways, and into each will be inserted a dinner-plate-sized LED red light. Some distance away, at a location with an unobstructed sightline to the taxiways and runways, a tower will be erected to host a fast-turning white radar bar that will scan the surface for planes and ground vehicles.

The result will be a prototype of the Runway Incursion Prevention through Situational Awareness system, or RIPSA, a network that would trigger the lights to turn red whenever a plane or vehicle is on the runway ahead, or if an aircraft is about to land on it. To achieve this, the radar will bounce radio waves off aircraft or vehicles on the surface, and the resulting detections will be transmitted to a computer somewhere on the airport grounds (no one involved in the project would say where). This computer also will receive GPS position reports from aircraft on the tarmac or just above it via FAA’s network of Automatic Dependent Surveillance-Broadcast antennas. In an instant, algorithms in the computer will decide which of the lights to click on to a bright red, visible both day and night, and which to turn off again.

The lights will be a failsafe on top of conventional radio communications with air traffic controllers. That should reduce the risk of a pilot making the potentially deadly mistake of turning onto or crossing a runway that’s occupied, for instance, by a plane that has started its takeoff roll. The stakes of such mistakes have been vividly illustrated by a series of near misses at U.S. airports in recent years. RIPSA, if it works as its developers hope, could have prevented many of them, such as a January 2023 incident at Daniel K. Inouye International Airport in Hawaii in which a United Airlines jet taxied across the runway where a Kamaka Air plane was landing.

For the RIPSA experiment, FAA issued a contract to Saab Inc., the Sweden-based supplier of air traffic management technology, in 2022, not knowing that a spate of close calls was about to begin. FAA would not grant me an interview, but officials could be hoping that the work will calm the nerves of the flying public and satisfy Jennifer Homendy, the chair of the U.S. National Transportation Safety Board.

“How many times are we going to have to issue the same recommendations over and over and over again?” Homendy said at an FAA safety summit in March 2023 in McLean, Virginia. She was referring to seven NTSB recommendations that haven’t been acted on, including one from 23 years ago: that FAA require all passenger plane airports to have technology to detect aircraft on the ground and prevent runway incursions by directly warning pilots. Homendy said FAA has rarely responded, but “we sometimes get the response that ‘It costs too much.’”

Cost is where RIPSA and the planned experiment in San Antonio come in. There are 450 commercial airports in the United States, but at the moment, only the 20 busiest of them seek to prevent collisions with the most sophisticated possible technology available: a combination of runway stop lights, multiple radars, antennas to triangulate locations by receiving transponder signals, and software to connect with the ADS-B network, plus algorithms to tie the detections into a surface traffic picture and display it for controllers. Twenty-three other airports have nearly all of that, minus the runway stoplights.

So, in San Antonio, researchers contracted by FAA aim to find out if all that is necessary.

“Can you drive that system with less accurate sensor data and still accurately turn lights on and off appropriately?” says James Kuchar of MIT’s Lincoln Laboratory outside Boston. Kuchar led development of the Runway Status Lights portion of the more sophisticated system and now heads the lab’s air traffic control technology efforts. “It’s taking the algorithms for a very complicated system and adapting them so they will work with lower-quality sensor data and see how well we can do.”

By complicated, he means, for example, ASDE-X, short for Airport Surface Detection Equipment Model-X, a package of equipment at 35 of the 43 airports that are equipped to address the risk of surface collisions. ASDE-X typically consists of 12 antennas to pick up the transponder signals, plus two kinds of radars and the ADS-B software, and displays showing controllers the shifting positions of planes and ground vehicles. There’s also software that emits audio and visual warnings for the controllers.

The remaining eight of the 43 airports have the ADS-B Airport Surface Surveillance Capability, which is similar to ASDE-X but also transmits the traffic information to the aircraft cockpits for display to the pilots. The busiest 20 of the 43 also have the Runway Status Lights network, which consists of lights on the runways that are triggered by algorithms that ingest the ASDE-X or ASSC sensor data.

The researchers have set out to see how close they can get to achieving the same level of safety with fewer lights, one kind of radar, ADS-B and without the controller or cockpit displays.

Specifically, at San Antonio, stoplights will be installed at six of the airport’s three dozen taxiway intersections with runways. That compares to Runway Status Lights, which calls for installing lights at intersections and also on the runways to hold planes from taking off: All told, 18 sets of lights at Boston’s Logan Airport and 34 sets for San Francisco, for example.

For the more sophisticated system, ASDE-X is a top cost driver because it depends on surface radars, which aren’t mass-produced, explains Christopher Oswald, head of safety and regulatory affairs for Airports Council International-North America in Washington, D.C.

“It’s not like you can go out and buy one off the shelf,” Oswald says. FAA did not respond when I asked for the cost of installing ASDE-X, the Runway Status Lights or both. Oswald said the cost of installing both at a single airport would run to tens of millions of dollars.

RIPSA, its backers say, could prove both the safety and business cases for more airports, because the costs of implementing and maintaining such a sensor and warning system should be much lower, although none would provide an estimate.

That’s the motivation for the San Antonio experiment, but safety will be paramount, and no one is rushing it. At the moment, researchers at Lincoln Lab are making the algorithm adaptions required to ingest the pared-back readings: from just the surface movement radar and ADS-B positions, says Wesley Olson, head of the Lincoln Lab group that develops aviation safety technology. A live feed will be established at the lab to receive ADS-B signals from planes in San Antonio that are taxiing, about to take off or just landed. The radar detections and ADS-B location information will be run through the RIPSA algorithms to evaluate whether they would have turned the lights on and off correctly, if they had been in place.

Once the algorithms and radar are in place, live testing won’t begin right away. Plans call for operating RIPSA in “shadow mode,” meaning the lights on the runways won’t be turned on and off, but the researchers will be able to view whether the lights would have done so correctly, Olson says. Once everyone is confident, the lights will start operating at the runways, although the researchers underscore that the goal here is not to replace the “hold short” or “cross” or “cleared” voice instructions from controllers.

If all goes as intended, the thinking is that safety will be enhanced, because now there would be voice and also a visual cue. Also, if an unexpected conflict occurred — a plane taking off on a runway without clearance, for example — the runway lights could signal a pilot to stop faster than a controller’s voice alert.

Two main findings will determine success for RIPSA, Kuchar says: how quickly the algorithms can determine that a light should be turned red, and how rarely false indications are given. Because ADS-B plus one radar won’t be as precise at locating planes and vehicles as the full sensor package for a Runway Status Lights airport, the researchers will have to adapt the RIPSA algorithms to cope with shifting estimates of aircraft movement. For RIPSA, they can tolerate this inaccuracy because the algorithms are only turning lights from off to red, not providing a display map for controllers as ASDE-X or ASSC must.

“Especially with these lower-quality surveillance sources, the targets are going to be jiggling around a little bit because it’s not perfect. And so some of those jiggles might cause the system to think the plane’s taking off when it really isn’t,” Kuchar says.

The jiggle issue is one reason why the ASDE-X and ASSC runway incursion warning displays for controllers won’t be part of RIPSA. Another reason is that these displays would be too expensive, Kuchar says.

“The Runway Status Lights algorithm can handle that maybe things are jumping around a little bit,” Kuchar says. “But a person looking at the screen is going to be really annoyed, seeing things jumping around or turning around. It’s not going to be an effective human interface.”

But adding warnings for the control tower for RIPSA may be an option to consider for the future, he says.

At San Antonio, researchers plan to seek feedback from pilots and air traffic controllers to see how pilots react to the lights and whether the red lights are effective, or not noticed enough, or adding too much to controllers’ workload and maybe even inducing new errors, Kuchar says.

“Part of this is very much human intensive, of actually getting people out there collecting data with real people using it, to see how it’s working,” Kuchar says. “We definitely cannot just design stuff in the laboratory and then just ship it out and get it up and running and expect it’s going to meet everybody’s needs the first time.”

If the RIPSA algorithms are proven accurate, the next question would be which airports could benefit most from the technology. To find out, a computer model of each airport must be built, and it must include the airport’s traffic patterns, its runway and taxiway layouts, the number of flights, and the number of flights that might be at risk of causing a runway incursion. RIPSA would then be wrapped into each airport’s model, and authorities could judge how well it would do at preventing runway accidents, Kuchar says.

RIPSA could become an option not only for smaller airports that don’t have Runway Status Lights, ASDE-X or ASSC but maybe even an option to replace the full array of sensors under ASDE-X, Olson says.

Also, during the San Antonio experiment, researchers working separately from real-world operations could test the effectiveness of RIPSA with only ADS-B, an option FAA is very interested in, Olson says, given the expense of purchasing and maintaining radars. The main challenge of relying solely on ADS-B would be how to account for airport vehicles that are not equipped with ADS-B transponders and the sliver of general aviation planes that may not either.

RIPSA’s success would ultimately be judged by whether it could extend some of the benefits of Runway Safety Lights, ASDE-X and ASSC to more airports. In her safety summit speech, Homendy of the NTSB pointed to ASDE-X and ASSC as effective but rare preventive measures for runway incursions.

“We have the technology. But ASDE-X is only at 35 airports. ASSC is at eight,” she said.

She also complimented the industry for a remarkable record of safety, while pleading for improvement from the spate of runway incursions in early 2023: “These recent incidents must serve as a wake-up call for every single one of us, before something more catastrophic occurs. Before lives are lost.”

About Keith Button

Keith has written for C4ISR Journal and Hedge Fund Alert, where he broke news of the 2007 Bear Stearns hedge fund blowup that kicked off the global credit crisis. He is based in New York.

Clarifying remarks

MIT’s Lincoln Lab is also researching a potential add-on to RIPSA, the Runway Incursion Prevention through Situational Awareness technology project, to address a problem that runway stoplights can’t solve: What if a pilot drives a plane onto an already occupied runway because of a misunderstanding with a controller, whose directions always supercede whatever the runway stoplights be showing?

The research, which was begun about three months ago, was inspired by a January 2023 incident at John F. Kennedy International Airport in New York. An American Airlines 777 took a wrong turn and crossed the runway that a Delta 737 had started to take off on. The captain of the 777 didn’t realize the mistake until the aircraft was in the middle of the runway and he saw the Runway Status Lights illuminated red. The airport’s ADSE-X sensor network alerted the control tower of the conflict, and a controller ordered the Delta pilot to abort the takeoff.

Natural language processing software could be applied to catch mistakes like that earlier, says Wesley Olson, head of the Lincoln Lab group that develops aviation safety technology. If the controller tells the pilot to follow a certain path via taxiways to reach a certain runway for take off, for example, and if the pilot reads back those directions incorrectly, the software would catch the mismatch. Or if the directions were read back correctly but the plane then took a different path than instructed, as determined by the plane’s ADS-B location transmission, the software would alert the controller or pilot. The technology could be applied to any kind of pilot-controller miscommunications, Olson says.

“Our intention in the end would be to extend that beyond just surface but also to accommodate issues in flight — detecting clearance issues in communications or detecting when pilots are not doing what they’ve been told to do a little bit earlier,” he says. — Keith Button

Runway collision
The last deadly U.S. runway collision was in 1991, when a Boeing 737 arriving at Los Angeles International Airport landed on the same runway where a SkyWest commuter plane was awaiting clearance to take off. The airliner crushed the smaller plane, killing all 12 aboard, as well as 22 of the 89 aboard the 737. Credit: AP/Scott Harms
Runway Status Lights at day and night
Day and night: Runway Status Lights like these are installed at 20 of the 450 commercial airports in the United States. For the San Antonio trials, lights identical to these will be installed so researchers can determine if a simplified version of the sensing system that triggers them can effectively avert ground collisions. Credit: FAA

A new light for safety