Inside Daedalean's push to certify machine learning avionics for the eVTOL era

Bas Gouverneur from Daedalean discusses the groundbreaking development of certifiable AI-based vision systems for aviation safety with Alex Scerri from Vertical Magazine.

The Zürich-based company, recently acquired by Destinus, has achieved a major breakthrough by developing certification methods for machine learning systems with both EASA and FAA approval. Their AI-powered technology enables aircraft to detect airborne threats, navigate without GPS, identify landing sites, and avoid wires using optical sensors and deterministic algorithms trained on vast datasets collected across diverse terrains.

• AI system operates independently without air-ground connectivity, using pre-trained algorithms loaded into aircraft avionics
• Successfully demonstrated 130-kilometer GPS-free flight using terrain-based visual navigation with Moog's Bell OH-58 helicopter
• Technology weighs only 4 kilograms with 200-watt power consumption, making it ideal for eVTOL applications
• Certification expected by end of 2025, marking the first ML-enabled certified avionics system in aviation history