Vehicle-Road Cooperative Positioning Algorithm Based on Road Code in GNSS-Denied Environment

In urban environments where signals from global navigation satellite system (GNSS) are vulnerable to interference or failure, including urban canyons, viaducts, and long-distance tunnels, the positioning precision of self-driving vehicles is significantly diminished due to the absence of global location information. The paper incorporates precise location information into road code, which can assist vehicles in achieving reliable, high-precision positioning. Based on the inertial odometer (vehicle-end) and road code (road-end), the vehicle-road cooperative positioning algorithm is proposed. First, the road code that contains sufficient location information is designed, and the regression-based one-stage road code detection network is developed. The network can detect road code accurately and in real-time, and then road-end location information can be obtained through decoding and pose estimation. An inertial odometer is used to acquire vehicle-end location measurements. Then an Error State Kalman Filter (ESKF) is designed to correct the vehicle-end location with that of the road-end and output high-precision location information. We conducted road tests to validate the performance of the proposed positioning system. The results demonstrate that this system can provide reliable and high-precision vehicle positioning without GNSS signals. The positioning precision can reach the decimeter level.