Daedalean, a startup developing software for vision-based flight control for traditional aviation and urban air mobility industry, announced today to have selected UNIGINE, the most advanced 3D engine developer for enterprise, as a base for creating its virtual simulation environment.
The simulation is used for training Neural Networks (NNs) for performing the functions of the "visual cortex" of autonomous aircraft or eVTOL (electric vertical take-off and landing vehicle) – a personal "flying car" of the near future.
Daedalean's product is aimed to become a certified autopilot solution based on advanced techniques in computer vision and machine learning. In July 2019 the startup received its second funding round with the VCs including Carthona Capital, Redalpine and Amino Capital, announcing that the funding raised, totaling $11,9 mln, would allow it to proceed along its roadmap to the final creation of an autopilot that outperforms human pilots in all of their functions. To create the product addressing the highest safety standards, Daedalean focuses on developing NNs that produce a predictable and explainable outcome, and providing adequate proof of the system's safety through a detailed analysis of the decision-making and functionality of its AI algorithm.
Such a system requires millions of hours of training to learn how to react to the most extreme and improbable situations that might occur during the flight; and the algorithms of visual recognition require as much training to achieve the "experience" ensuring their "knowledge" of any imaginable landscape below and around the flying vehicle. They should be able to detect and recognise any kind of obstacle in urban and rural environments that prevents safe take-off, flight and safe landing: wires, masts or trees, animals or statues, cars or bushes. And the opposite task – they should be able to identify a safe zone for emergency landing, distinguishing swamps and fields, streets and roofs, sands and ices, in any geographical environment and any landscape.
Achieving this amount of training hours is highly impractical in real-life, due to the sheer amount of time and the need to recreate edge-scenarios thousands of times. To this end, Daedalean approached UNIGINE, a provider of enterprise-grade real-time 3D solutions. UNIGINE has vast experience of working with aerospace giants, including companies and organizations such as RSC Energia, the German Aerospace Center and Korean Aerospace Industries. The UNIGINE platform is perfect for simulating aerial flight, owing to its extremely precise coordinate system, uncompromising level of detail even at the horizon, lightning fast object loading and perfectly realistic atmospheric model.
UNIGINE provides a unique version of its software development kit, tailored specifically for applications in aerospace and ground transport simulation. Unlike most other engines, it uses a realistic "ellipsoid" representation of the Earth, allowing aerospace users to correctly simulate geodetic or "Great Circle" trajectories. Its advanced projection tools allow to effortlessly initialize multi-projector or dome screen setups, in addition to supporting a variety of industry-leading VR headsets – including the enterprise-grade Varjo VR-1 headset. Native support of thermal and night vision complements the picture.
These features combine with top-notch rendering quality and advanced scripting capabilities including APIs for C++, C# and UnigineScript, to make UNIGINE one of the best 3D development tools available on the market – which is why Daedalean chose it for its crucial task of training the pilots of our future.
Daedalean CEO and Founder, Luuk van Dijk, commented on the contract: "Testing our AI-based situational awareness guidance software is of vital importance for obtaining full autonomy – a key enabler for the scalable Urban Air Mobility. After an extensive trial phase we chose UNIGINE because it provides the high-quality environment for training our neural networks."
Denis Shergin, CEO of UNIGINE, also commented on the news.
"Usually our engine is used to build simulators for real human pilots, but it works just as well for training artificial neural networks. We at UNIGINE are happy knowing that our work could be instrumental in bringing to us the future of urban transportation, and will support Daedalean in anything they might need to fine-tune their simulation software."
|Contact details from our directory:|
|Daedalean AG||Autopilots, Terrain Awareness and Warning Systems|
|UNIGINE LLC||Computer-aided Simulation|
|Related directory sectors:|
|Flight and Data Management|