A team in BAE Systems Australia has been working with the Defence Science and Technology Group in the Department of Defence, as well as the universities of Melbourne and Adelaide, to help solve this problem by augmenting sensor data from the environment such as real-time LiDAR (Light Detection And Ranging), electro optical and radar sensor data; with historic satellite data.
Andrew, the Software and Algorithms Lead, talked about the purpose of the programme: “We want to provide autonomous vehicles that can effectively achieve their goals, which can only happen when vehicles are able to identify what’s around them and behave accordingly - they need to know if it’s a bush in front of them they can drive through.”
The team has made some significant progress in trials, demonstrating that the vehicle can drive around, take in raw sensor data, compare it with the satellite data and fuse it together to make a better decision about objects than if it was relying on either data source in isolation. The vehicles were able to use this to avoid obstacles and fed back ‘semantic maps’ to operators, sorting the objects around them into groups with a high level of accuracy.
We want to provide autonomous vehicles that can effectively achieve their goals, which can only happen when vehicles are able to identify what’s around them and behave accordingly - they need to know if it’s a bush in front of them they can drive through.Andrew, Software and Algorithms Lead
Many autonomous systems use Simultaneous Localisation and Mapping (SLAM), where you build a map as you explore an area, but we have added geo references gathered from the satellite data to add accuracy. We’re also looking at whether we can combine SLAM with other contextual information to enhance the ability to cope with GPS denials and spoofing.
The team is currently working through a four-year programme, but has demonstrated many successes already. Last year they were able to show that a team of vehicles could share awareness of the local area, allowing them to navigate successfully even when sensors were obscured. They’ve also been able to show that when there’s a discrepancy between sensor and satellite data, the vehicle is able to give priority appropriately - re-routing to drive around buildings that had been built since the satellite data was created, for instance.
The next step in the programme is to demonstrate these vehicles with a full-scale 6-wheeler. There are also plans to add Electronic Warfare (EW) capabilities, both to detect sources of EW and use it in the prosecution of its mission.