Independent aerospace startup Daedalean has published a roadmap outlining its path to developing a self-flying aircraft by 2028.
The Swiss company is working on a flight system completely independent of vehicles, starting with situational awareness software and sensors that will be used as assistance to pilots. The company ultimately aims to be able to convert any type of aircraft to fly completely autonomously — without the assistance of an in-flight safety pilot or a remote pilot on the ground — using artificial intelligence (AI) and machine learning.
“If you really want to replace the pilot, whether it’s remotely or on the plane, you have to create systems that can do that,” said Daedalean CEO Luuk van Dijk. FutureFlight. “So, for that, you have to have this complete awareness of the situation, and then, you need enough sensors, and then you need these parts of the AI to understand those sensors.”
To start, Daedalean is working to achieve General Aviation Situation Awareness System certification, under Design Assurance Level (DAL) C, with the goal of eventually certifying a fully autonomous flight system up to DAL A, the highest level of Design Assurance. that can be applied to airborne software. The company plans to launch Eye of the pilot Experimental machine learning-based pilot assistance for general aviation in 2023. It is developing the system in collaboration with Florida-based avionics company Avidyne.
“Our partner Avidyne makes the crates, we go through certification training and when that is done, we will have a supplemental type certification,” Van Dyck said. “This will be a huge moment for the industry as a whole because it will be the first time a machine learning component has been certified for Design Assurance Level C, which means it is suitable for safety situations with significant impacts.”
The first application of Daedalean technologies is based on computer vision. Daedalan wrote in a newspaper White papers It describes its roadmap to 2028.
“By working on this, we solved the two toughest tasks: first, creating a machine learning-based technology capable of recognizing, classifying, and interpreting sensor inputs at true flight distances, velocities, and uncertainties, and working in a computing box of reasonable size, weight, and power consumption. Second, to establish the principles that On the basis of which this technology can be adopted, working alongside regulators to develop ways to ensure design for it,” the paper adds.
The roadmap then describes what Daedalian calls the IXS, or “System of Everything,” which will use data from cameras and other traditional tools — such as GPS, altimeters, lidars, radars, and traffic information sources — to get you to work Meteorological conditions tool. “The Daedalean IXS fully certified road and traffic collision avoidance, as well as landing guidance and flight plan following, will provide capabilities, at VMC and IMC, certified to appropriate levels to operate with only minimal crew supervision (EASA Tier 3),” states .
Daedalean plans to further enhance the IXS with vehicle-to-vehicle communication capabilities, allowing aircraft to share map data and flight plans to coordinate secure routing. This feature, called XXS, may be particularly useful for air traffic management in urban air mobility, according to the paper.
To obtain these new technologies based on artificial intelligence and machine learning, Daedalean is working closely with the European Union Aviation Safety Agency (EASA). In February 2020, that agency published a file artificial intelligence roadmapwhich formalized the certification phases of machine learning applications.
EASA’s AI Roadmap is divided into three phases. The first phase, which runs until 2024, is an exploratory period. The next phase, from 2024 to 2028, is called “Framework Standardization,” which is when the agency plans to release its first approvals for AI and machine learning components. The third phase, called “Pushing the Barriers,” will see further innovation in AI technology and the first fully autonomous commercial airlifts.
Daedalean also participated in a joint research project with the US Federal Aviation Administration (FAA) in Neural network-based runway landingwhich can help shape agency policies on artificial intelligence and machine learning.
To obtain fully autonomous flight technology certification, Dedalian expects regulators to require the company to complete several years and several thousand hours of flight and service.
“Technically, we haven’t tested the flights yet because we don’t have the qualification units from Avidyne,” Van Dyck said. FutureFlight. “But for software purposes, we collect data and test that our algorithms work in flight without being tied to the aircraft. So this is a feature of situational awareness. You can test it on recordings.”
You can read Daedalean’s full white paper, as well as several handouts describing their standalone systems in detail, on the company’s website. website.