Beyond Concorde: FOD detection in Twenty-First Century

  • Published
  • By William Giannetti

Twenty-one years after Concorde flights tragically ended, universities and private industry are pioneering artificially intelligent technologies that detect foreign object debris (FOD). These advancements, partly born from the disaster, highlight crucial optimization opportunities with the potential to enhance flight safety for the U.S. military.

The Concorde Tragedy

            On July 25, 2000, at 12:30 p.m., 100 men, women, and children checked in for Air France Flight 4590 at Charles de Gaulle Airport bound for New York City. On the jetway was the airliner Concorde. Distinctively delta-winged and luxurious, it could travel from Paris to John F. Kennedy Airport in less than four hours. By 1:30 p.m., the passengers and crew had settled in for their intercontinental trip. Captain Christian Marty, a seasoned pilot since 1967 with over 13,000 hours aboard various commercial aircraft, including 317 aboard the Concorde, was at the controls. With 40,000 flights and 900,000 flying hours over the aircraft’s life span of 32 years, the Concorde fleet had experienced no fatalities.

At 4:40 p.m., Marty steered Concorde and accelerated down the runway at 200 miles per hour. Suddenly, the tower’s controllers saw flames erupting from the massive engines on the left side. Sixty-five hundred feet into its taxi, and with 1,000 feet of runway to go, Marty committed to the takeoff, though it was too late. The damage was catastrophic as the jet crashed into a nearby hotel, killing four people, nine crew members, and all 100 passengers.1

Investigators attributed the horrific accident to foreign object debris (FOD). An exhaustive five-week inquiry revealed the object’s origin and the role it played. A Continental Airlines DC-10 had taken off five minutes earlier, leaving behind a strip of perforated metal from one of its engines. Approximately 1.5 feet long, the object punctured the Concorde’s front tire during takeoff. A chunk of rubber tumbled through the ill-fated airplane’s fuel tank, leading to the deadly chain reaction.

The era of supersonic air travel came to an end. Air France ceased Concorde’s flights in May 2003, and British Airways retired its fleet on October 24, 2003. The incident, decades later, still famously underscores FOD detection systems’ criticality in aviation safety, the importance of preventing runway debris, and its potentially devastating consequences.2

The “FOD Walk”

In aviation history, no tradition is as time-honored, and as tedious, as the FOD walk. This brief use case will discuss some solutions small technology companies and universities offer to reduce what is a time-intensive process.

In a worldwide estimate, the Federal Aviation Administration (FAA) calculated FOD damage costs up to $22.7 billion per year.3 Prices for engine fan blade replacements often exceed $42,000.4 To minimize damage, Airmen and Sailors aboard aircraft carriers engage in FOD walks that detect debris as small as a pebble, a common washer, or a rivet. Early in the morning, or at dusk before operations wind down for the day, the troops walk a flightline, twelve abreast. Long overdue advances in automation, however, afford military personnel a return to their primary responsibilities—controlling and managing airstrip operations.

First, many automated FOD systems, like at Boston Logan and Seattle-Tacoma (Sea-Tac) international airports, are ground-based. One company, XSight Systems, integrates a millimeter wave radar into runway edge lights. The radar is omnidirectional and works between the 30 to 300 GHz range. It can detect objects a few millimeters in size, such as screws or bits of rubber from tires. Officials at Boston Logan say XSight sensor installation costs $1.7 million, with around half covered by the FAA. In 2016, SeaTac reported that the sensors picked up 8,256 objects, with 123 deemed potentially harmful. The system is available 24 hours a day, and viewers witness the results via laptop or tablet before walking out to investigate.5

The Air Force, in partnership with Siemens and TurbineOne, is developing a four-wheeled robot known as an “FOD dog.” Powered by a small electric motor, the compact rugged vehicle has a wheeled chassis with off-the-shelf scanners on its front, left, and right sides. The nature of today’s graphical processor units (GPUs) makes the device’s effectiveness possible, as TurbineOne’s CEO Ian Kalin explains: “Machine learning models have gone through waves of ‘miniaturization,’ which basically means that code that used to require large computer resources and many gigabytes can now operate with only megabytes worth of space.”6 The scanners, sensitive to the tarmac’s surface abrasions, outperform standard cameras in low visibility conditions such as heavy rain or light snow. Vancouver, Washington-based Maren-go Solutions has also developed what CEO Chris Thobaben calls a “runway Roomba,” which, like the FOD dog, can detect objects two centimeters long. Based on trials at Joint Base McGuire-Dix-Lakehurst, New Jersey, Thobaben’s 330-pound vehicle boasts FOD walk reduction times from 43 to 3 hours.7

But distinguishing between regular runway features, such as small cracks, textures like painted lines, and foreign objects, must be clarified for a detection system. Taking a layered approach involving the air is a better course of action. Synthetic aperture radars (SAR), strategically positioned on light poles, transmit microwaves that capture a metallic object’s reflection, regardless of most lighting conditions and time of day. Optical cameras on board quadcopter drones could also aid in correlation, object detection, and identification. One model, developed by the University of Pennsylvania, is equipped with an “eagle claw” that swoops in and retrieves objects.8

Drones, however, carry a stigma and certain federal restrictions. If a malfunction occurs, a small quadcopter like the MD4-1000, for example, risks becoming FOD itself. FAA regulations forbid private drone flights within five nautical miles of an airport. Operations near military installations are more stringent. To remedy this, University of California at Berkeley students recommend a “cross-departmental” approach where air traffic controllers work in close coordination with a three-person team of drone operators. An overseer for the pilot and copilot would identify potential hazards and ensure a safe flight.9

Putting It All Together

China’s Nanjing University lends a practical working model for semi-autonomous FOD detection.10 Once the system is powered on, staff members take their stations as the software conducts a self-check. Millimeter wave and SAR sensors sweep the airstrip. If the sensors discover a foreign object, a monitor will receive the information and decide to investigate. From there, a teleoperated FOD dog confirms the object and transmits its location to a control room, where a removal team stands ready. Another method of detection practiced at a northern China airfield could also be adapted for American purposes. People’s Liberation Army Air Force rapid platoons use Segway transports outfitted with suction devices that vacuum FOD from crevices.11  

An independent closed system is a more secure option for FOD detection. A local network integrates radar, drones, and SAR, ensuring data remains within a virtual perimeter with end-to-end encryption and local data processing on GPUs. Detection algorithms are being perfected. While some can identify straight objects like the Concorde incident’s metal piece, curved debris pose complexities. Nanjing University suggests a computer vision technique called Hough Transform line detection for irregular shapes, which dices images into pixels. Engineers could also establish a local convolutional neural network, which essentially acts like a detective with a magnifying glass observing the airfield and all the objects surrounding it. This neural network first zooms in, examining the pixelated debris one frame at a time; then it studies corners, sharp edges, and other distinguishing features. As it zooms out, the program reassembles the entire picture, memorizing it in great detail.

Since the Concorde tragedy, technological strides in industry and academia have put a comprehensive artificial intelligence-driven FOD detection system within reach. Yet challenges persist in curbing false positives and navigating federal regulations. Nonetheless, by refining algorithms and adopting streamlined operations, the Air Force, with some outside assistance, can revolutionize aviation safety. Its advancement in this area, minimizing human intervention, promises not only to conserve vital resources but also to save lives as well.


Lieutenant Colonel William Giannetti, USAFR, is the 62d Airlift Wing’s reserve senior intelligence officer and a consultant based in western Washington.

1 “Air France Flight 4590 FOD Explained,” FOD Prevention (website), December 6, 2016,

2 Ministry of Transport Equipment and Housing, General Inspector of Civil Aviation and Meteorology, “Accident on 25 July 2000 at ‘La Patte D’Oie’ in Gonesse (95) to the Concorde Register F-BTSC Operated by Air France,” translated preliminary report, October 10, 2001,

3 B.G Hilburn and B.S Pesmen, Foreign Object Debris Detection System Cost Benefit Analysis, Research Report no. DOT/FAA/TC-22/47 (Atlantic City, NJ: Federal Aviation Administration, May 2023),

4 Mike W. Ray, “New Fan Blade Contract Saves Air Force Millions,” Tinker Air Force Base (website), February 24, 2012,

5 “Port of Seattle, Seattle Tacoma International Airport (SEA) Case Study” (Waltham, MA: XSight Systems, May 28, 2017),

6 David Hambling, “U.S. Air Force Developing Robot ‘Dogs’ to Keep Runways Clear of Hazardous Debris,” Forbes, January 10, 2022,

7 Author’s notes from National Security Innovation Network (NSIN) Conference, Seattle, WA, November 9, 2022.

8 Justin Thomas et al., GRASP Lab, University of Pennsylvania, “Avian Inspired Grasping for Quad Rotor Micro UAVs,” February 7, 2013, YouTube video, 0:39,

9 Hiromichi Yamamoto et al., Drone-enabled Foreign Object Debris (FOD) Removal System in ad hoc Situations” (grant submission by the University of California, Berkeley, Virginia Space Grant Consortium’s Airport Operation and Maintenance Design Challenge, 2016),

10 Li Ang, “Research and Design of an Airfield Runway FOD Detection System Based on WSN,” International Journal of Distributed Sensor Networks 2013, published online December 16, 2013,

11 Zhu Bailian et al., “The Chinese Air Force Has Developed a Sharp Weapon for Individual Patrol Runway to Prevent Foreign Objects from Damaging the Engine,” Sina News, February 24, 2016,