Iain Clarke discusses how expanding the operational domain of localisation systems in autonomous vehicles can enable more reliable navigation
Autonomous vehicles are revolutionising the transport industry, from personal mobility with robotaxis, to freight transport with autonomous trucks and HGVs and last-mile delivery vehicles in-between. Whilst autonomy promises the benefits of increased efficiency, long-term cost savings, and continuous operation, there are still a number of technical challenges to overcome.
One of these challenges is how to ensure reliable operation, with appropriate accuracy, in all potential environments. Oxford Technical Solutions (OxTS) is working on enabling autonomy in more and more environments with the Generic Aiding Data (GAD) interface, which streamlines the integration of sensors for use in the navigation engine, overcoming the limits of traditional localisation systems.
The scope of potential environments the autonomous vehicle will encounter is known as the Operation Design Domain (ODD). The ODD is a key specification in the design of an autonomous vehicle, and clearly, the smaller and more restricted the ODD, the easier it is to achieve. However, in order to start reaping the benefits of autonomy and maximise its usage, the ODD needs to expand so it can perform on par with a manually operated vehicle. For the localisation system, which positions the vehicle in the world and in relation to its surroundings, Global Navigation Satellite System (GNSS) reception is the biggest constraint in maintaining a highly accurate position measurement. If the ODD contains inner-city driving like a delivery robot might experience or long tunnels in a freight transport route, satellite visibility is going to be obstructed, and the GNSS position accuracy will be degraded.
To counteract the weaknesses of GNSS, additional sensor data can be used to provide alternative sources of information. For example, Ultra-Wide Band (UWB) beacons can provide positioning in the absence of GNSS, a wheel speed signal from a vehicle can provide an independent source of velocity, or LiDAR data can be used for Simultaneous Localisation and Mapping (SLAM). With the OxTS GAD interface, integrating external sensor data is streamlined and opens possibilities for new and novel sensors. The external data gets integrated in the Extended Kalman Filter alongside the inertial data to generate an improved overall localisation solution that is no longer bound by the requirements of GNSS.
The integration of external sensors through the GAD interface in combination with OxTS inertial navigation systems (INS) is enabling reliable navigation and localisation in more and more environments, opening the doors for autonomous vehicle designers and transport service providers to widen their operations and expand autonomy to new and exciting areas.
Iain Clarkeis the senior enginer at OxTS