The technology developed by Delfox consists in giving objectives to the AI, which must then find a way to achieve them. When it comes to AI, it is essential to understand that this intelligence is based above all on learning.

It is therefore learning mechanisms that lie at the heart of Delfox’s development, which has progressed significantly for over two years in cuttingedge skills like deep learning and reinforcement learning, as well a s the related advanced algorithms.

The goal is to teach a machine to react autonomously, without indicating how to resolve an issue. The machine itself proposes solutions, which will lead to rewards or penalties; it will therefore learn from its mistakes.

For example, teaching a drone to go from point A to point B does not mean telling it to avoid collisions or to accelerate at certain points of the journey; it is about letting it react by itself and rewarding or penalising it based on the solutions it proposes. Potential applications are vast.

There is, of course, the area of satellites, in which Delfox is already working with Ariane Group for space surveillance purposes. Delfox participates in detecting satellite trajectories based on data provided by the GEOTracker space surveillance network to avoid collisions and interference.

But the fields of application are a lot more extensive than just satellite uses: autonomous military and urban drones, cars, logistics, defence, the navy, and more are all potential areas of interest.

Autonomy will no doubt be a key segment of activity in the next decade, and Delfox is already one of the most successful players in the field. With a team of 15 people, Delfox aims to reach €1m in revenues in 2021 and is already working with Ariane Group, Dassault Aviation, Thales and the DGA (French government defence procurement and technology agency).