Low-latency GNSS multipath simulator for real-time applications in autonomous driving

It is essential to periodically provide precise position estimations to realize fully autonomous driving in all kinds of environments, especially in dense urban areas. Global Navigation Satellite System (GNSS) technology has the potential to provide positioning accuracy at a centimetre level. In urban environments the occurrence of satellite signal reflection and diffraction, also called multipath, is a challenging source of GNSS errors. A scalable real-time multipath simulator is created to face this problem and mitigate potential multipath errors. The inputs of the simulator are 3D building data, satellite positions and vehicle positions. A GPU memory manager and a specialised load balancer allow fast multipath analysis. Experiments show that the outputs of the simulator can improve the positioning accuracy of the positioning engine. The scalability of the simulator is shown by connecting the multipath simulator with a traffic simulator.