Hensoldt completes second data-gathering campaign for AI training database

Hensoldt's technology demonstrator equipped with various sensors. (Hensoldt)

German defence company Hensoldt has completed a second measurement campaign to collect data for a testing and training database for artificial intelligence (AI) algorithms.

The campaign is part of a European research and development project called STORE (Shared Database for Optronics image Recognition and Evaluation). As part of STORE, the foundation for a common European database for testing and training AI was created during the first measurement campaign in Fontevraud, France. STORE was initiated by, and receives funding from, the European Defence Fund (EDF).

In the second campaign, eight partners out of a total of 21 were involved and it took place in Kongsberg, Norway. “Its [Norway's] rugged landscapes and occasionally extreme weather conditions provided the ideal testing ground for AI algorithms,” according to the information on the STORE website. The organisation sees the country's terrain as key in training AI systems to adapt to diverse environments and provide accurate analysis.

A Hensoldt technology demonstrator, equipped with different sensor systems including the See Through Armour System (SETAS) 360°, was used during the campaign for reconnaissance, the company said in an 8 May press release. The electro-optical observation system for armoured vehicles is for 24/7 close-area observation with a 360° observation angle around the vehicle. Its visual sensors can recognise a person at more than 300 m, according to the company.

The campaign's collected data include recordings of different sample infantry vehicles, unmanned aerial vehicles (UAVs), infantry, and other personnel throughout the day, night, in snowy and cloudy conditions, as well as fog.