A team of researchers at the University of Cambridge has developed a solution to help air traffic controllers determine whether a drone is about to fly into a restricted area, based on surveillance data and algorithmic estimates. The technology can also be used to control robots and self-driving cars.
According to the naysayers, every new technology causes exactly as many problems as it solves, and in the case of drones there is undeniably some truth in this, because after the initial “honeymoon period” we have seen more and more reports of their misuse. The most controversial case was the one at London’s Gatwick Airport in December 2018, where large industrial drones flew into the runway area and disrupted traffic for days. The disruption caused by the drones affected hundreds of flights and disrupted the lives of 140,000 passengers.
Researchers at the University of Cambridge have been looking for a cure for problems like these, and have developed a solution that combines statistical techniques with radar data to predict the flight path of drones and tell whether they will fly into the closed area. The technique is a huge help to air traffic controllers because it can spot the nimble little flying objects before they enter the exclusion zone and indicate whether they pose a threat to other aircraft. This predictive capability, integrated into an automated decision support system, can relieve the burden on drone surveillance monitors, who then only have to deal with drones that pose a real threat.
The use of drones has become commonplace in recent years in agriculture, e-commerce and many other areas. But as they have spread, so have the security risks, especially with the advent of low-cost, yet powerful devices that are accessible to all.
At Christmas 2018, drones caused a huge commotion at Gatwick Airport, and the incident – the perpetrators of which, incidentally, have still not been found – has brought to the attention of the whole world the dangers of these flying devices when used with malicious intent or simply without proper preparation. While there is of course a law enforcement aspect to the problem – careful regulation and prosecution of offenders to force drone owners to use drones as required – researchers at the University of Cambridge have taken a different approach, seeking to increase the intelligence of drone-detecting surveillance systems.
Conventional aircraft are required to report their position every few minutes to avoid collisions, but drones have no such requirement. However, some form of automated air traffic control is also needed for drones, the researchers say. However, the problem is that, unlike large, fast-flying, cumbersome passenger aircraft, drones are small and agile, making them harder to track and easy to mistake for birds.
Potential threats should be detected as soon as possible, but their appearance should not be overlooked, as stopping air traffic is a drastic step that can cause huge damage and is particularly unforgivable if it is caused by a flash in the pan.
There are several possible methods for monitoring the areas around airports: a typical drone surveillance system consists of a combination of different sensors (radar, radio frequency sensors, cameras). A drone is considered a threat if it is suspected of intruding into the restricted area or if it shows an irregular movement pattern. For safety reasons, it is not possible to shoot down or otherwise disable a drone at an airport, and the only way to avoid tragedies is to close the airport completely. Researchers wanted to create a tool that would ensure that this last resort would only be used when justified.
The software-based solution uses a stochastic model to determine the drone’s flight intentions. Most drones navigate using waypoints, i.e. they fly from one point to another, and the entire route is made up of several such points. In the tests, based on real radar data, the Cambridge solution was able to identify the drone and predict in real time whether it would reach the next waypoint based on the vehicle’s trajectory, speed and other data. The new technology detected potential threats in seconds, giving air traffic controllers enough time to decide what action to take.
Finally, one more thing that makes the Cambridge team’s results particularly remarkable: the algorithm they have used can be applied to other fields (maritime safety, robotics, self-driving cars, etc.).