Recohere: Verification of an AI-native platform for data-driven digital twins in the defense sector

Important results from the project

The technical goals were largely met. The platform was productised into an MVP delivered on-premise, and the focus shifted early from vehicle sensors to drone data, the most likely data source for military use. Using drone data from WARA PS we built our first photorealistic, testable 3D world of real Swedish terrain. What we did not reach was a military end user with a defined need, which is required to move into an operational environment.

Expected long term effects

The ability to turn a short drone flight into a photorealistic, testable 3D world of real terrain can in the longer term support situational awareness and mission planning, simulation and training in environments that are hard to train in for real, and test and validation of autonomous systems such as unmanned ground vehicles and drones. All processing runs inside the customer´s own environment, which is a prerequisite for sensitive data.

Approach and implementation

The project was carried out by Recohere´s own team in four work packages: architecture and requirements, technical verification against civil reference data, verification against a defence-relevant use case, and strategic packaging. The only major deviation was the shift to drone data as the primary data source, made early and without delays. The project was extended by one month to cover the programme´s final conference.