Autonomous flight is just over the horizon, with multiple companies working on pilotless aircraft. But Daedalean is a key company in attempting to bring artificial intelligence into general aviation. As the company’s first product nears certification, we take a look at the steps in reaching full autonomy, with a few words from the company’s CEO.
The three main parts of autonomous flying
To create a system capable of adequately flying a full flight, the Daedalean team identified three main challenges to overcome. The first, positioning and navigation (the system can correctly identify where it is in relation to its surroundings, without the use of a GPS). When a human is flying a plane, if the visual conditions allow it, when the GPS fails they can identify landmarks by sight and position themselves according to what they see. On principle, the Daedalean system also works like this using data from cameras mounted to the aircraft.
Next, traffic detection and tracking (the system must find other aircraft within range, as well as other flying hazards such as birds). This is done without the use of ADS-B, which is one of the sources used for tracking aircraft on Flightradar24, and instead uses the cameras.
And finally, landing guidance. For this, the system has to recognize runways and helipads, as well as emergency landing sites, leading to it being able to land safely. This again, is done through the use of recognizing different aspects of images which are fed from the camera into the machine.
In the short video above, you can see the camera’s intake on the left hand side, and on the right how it manages to use what it sees to plot itself on a map.
The stages to autonomy
“What we are doing eventually is full autonomy… But it’s a long way ahead yet, ” says the company’s head of marketing, Olga Sternik. It’s highly unlikely that we will see aircraft which can comfortably fly without any input from a human in the foreseeable future, however what we will begin to see is the gradual phasing in of the computer’s control. The first stage in this process, is to have the computer give instructions to the pilot. Here, the AI (artificial intelligence) acts as a co-pilot, monitoring surroundings and assisting with foreseeing risks, whilst the pilot takes the actions. The AI is then given slightly more control, now being able to execute some tasks, however the pilot still makes the decisions and has the overriding power to intervene whenever necessary. The goal for Deadalean is not to necessarily totally remove all humans from the flight deck, but rather the human’s role changes to monitoring rather than doing. In the same way that commercial pilots still have a multitude of monitoring tasks when autopilot is activated.
Communication
Communication is vital whilst flying, and this next stage is where the AI starts to turn from auto-pilot to just pilot. Daedalean’s CEO, Luuk Van Dijk also sees room for AI in ATC operations, “due to its clear efficiency benefits, AI in the world of ATC (air traffic control) has certainly been proposed before, but it’s unlikely due to it possibly resulting in a loss of jobs,” he says.
However, communication from the Daedalean system will be automated, which Van Dijk argues brings clarity to the cockpit. He points to a hypothetical situation in which two pilots may disagree on what the ATC has instructed them to do, however the AI would be able to recognize this clearly. Developments on the AI’s ability to listen effectively are underway, with data from sources such as popular website LiveATC.net being used to train it to identify different aspects of aerial communication such as call signs and basic commands like inputting squawk codes.
Safety whilst flying
Many people have a fear of flying, and a common concern is that autonomous flight will add to the danger and increase that fear. For Daedalean, the safety benefits are very much clear. Daedalean are bringing autonomy into General aviation as their primary market, and Luuk says that General aviation is a sector “where you can clearly add safety.” This is something that is easily noticed by the number of emergencies from small aircraft which are seen on Flightradar24. Here, there is often only one pilot, and so if something were to happen to them, the full system could take control and keep flying. Equally, there is more speed when operating under emergency circumstances, as well as the fact that the AI will have monitored each element beforehand and will understand what has gone wrong and use it as a learning opportunity. As always, a set of rules will be put into place to ensure safety when flying autonomously, known as digital flight rules (DFR). The concept was first devised by NASA and is similar to Instrument flight rules (IFR) and Visual flight rules (VFR).
In this video, the AI is recognizing a runway for the aircraft to land on. With enough development, the system could safely land the plane on a runway should something happen to the pilot.
Piloteye™
The first product Daedalean is releasing is Piloteye™, which is in coordination with cockpit display manufacturers Avidyne. It is currently progressing through certification, a unique process as a product that uses machine learning has never been certified before. This has led to a joint effort from the FAA and EASA certifying it together. There is little competition at the moment for Daedalean; “I think some people are waiting for us to be the canary,” says Luuk. “People are making noises about things they want to do, but at the moment we are a pathfinder.” The new product will be able to identify other aircraft in the surroundings and place it on the display, and is expected to reach the first few stages of certification within the foreseeable future.
After the release of Piloteye™, this area of the industry is expected to grow, especially with large civil aviation manufacturers taking interest in Daedalean. As of November 29 2023, Daedalean recently announced a partnership with Xwing, another leading autonomous aviation company. This should further enhance the companies’ research into pilotless flight. Although Piloteye™ isn’t full autonomy like some might have hoped, it is the first step of many to integrate AI into the future of aviation.
![An example full flight summary response from the Flightradar24 API: { "data": [ { "fr24_id": "0987654321", "flight": "SK1415", "callsign": "SAS1415", "operated_as": "SAS", "painted_as": "SAS", "type": "A20N", "reg": "SE-DOY", "orig_icao": "ESSA", "orig_iata": "ARN", "datetime_takeoff": "2023-01-27T05:15:22", "runway_takeoff": "12R", "dest_icao": "EKCH", "dest_iata": "CPH", "dest_icao_actual": "EPWA", "dest_iata_actual": "WAW", "datetime_landed": "2023-01-27T06:15:10", "runway_landed": "27L", "flight_time": 3600, "actual_distance": 1007.74, "circle_distance": 6245, "category": "Passenger", "hex": "4A91F9", "first_seen": "2023-01-27T05:06:22", "last_seen": "2023-01-27T06:18:10", "flight_ended": "true" } ] }](https://www.flightradar24.com/blog/wp-content/uploads/2025/04/Flight-summary-API-cover-300x148.jpg)
![An example full flight summary response from the Flightradar24 API: { "data": [ { "fr24_id": "0987654321", "flight": "SK1415", "callsign": "SAS1415", "operated_as": "SAS", "painted_as": "SAS", "type": "A20N", "reg": "SE-DOY", "orig_icao": "ESSA", "orig_iata": "ARN", "datetime_takeoff": "2023-01-27T05:15:22", "runway_takeoff": "12R", "dest_icao": "EKCH", "dest_iata": "CPH", "dest_icao_actual": "EPWA", "dest_iata_actual": "WAW", "datetime_landed": "2023-01-27T06:15:10", "runway_landed": "27L", "flight_time": 3600, "actual_distance": 1007.74, "circle_distance": 6245, "category": "Passenger", "hex": "4A91F9", "first_seen": "2023-01-27T05:06:22", "last_seen": "2023-01-27T06:18:10", "flight_ended": "true" } ] }](https://www.flightradar24.com/blog/wp-content/uploads/2025/04/Flight-summary-API-cover-768x378.jpg)
