Case Studies

GPS-denied navigation UxVs

Technologies

Hybrid modelling
Real-time AI
Edge computing

Client’s Need

Unmanned vehicles (UxVs) must be cost-effective while retaining sufficient positioning accuracy to perform long range and long duration missions for maintenance, reconnaissance and defence purposes. In the absence of reliable global positioning information, navigation is carried out in dead reckoning which currently can achieve satisfactory positional accuracy only when using bulky and expensive sensors.

Approach

We are carrying out an integrated approach, intervening both at the sensor level and at the system level: we are developing data-driven algorithms to characterise and correct sensor noise, thus providing better measurements with cheaper sensors, and we are working on innovative AI-powered data fusion algorithms that use all data available on-board to estimate the UxV’s position through time.

Main Outcomes

Our integrated approach has shown excellent performance in simulation and is now being deployed on the edge and tested in controlled environments.