Difficult GNSS environments and adverse weather conditions require a fusion of many sensors to maintain lane-level accuracy for autonomous platforms, without incurring high costs that would inhibit widespread adoption. Radars are an attractive option in a multi-sensor integration scheme due to being robust to adverse weather and insensitive to lighting variations.
The multi-radar integrated version of AUTO uses inertial navigation, real-time kinematic GNSS, odometer, and multiple radar sensors with high-definition maps in a tight non-linear integration scheme. AUTO can reliably produce accurate and high-rate navigation outputs in real-time and under all urban environments. Key performance indices during simulated GNSS outages quantify the accuracy of the solution for prolonged periods.
Read the full paper here.