Low-cost vector map assisted navigation strategy for autonomous vehicle

Differential Global Positioning System (DGPS) is a crucial component in the perception system of todays autonomous vehicle (AV). Traditional navigation technique takes advantage of the accurate positioning of DGPS and high-definition (HD) map to facilitate path tracking. However, HD map occupies large storage space due to its extensive information on the detailed structure of the environment, which is extremely costly to deploy onto in-vehicle computing devices. This work proposes a low-cost vector map based navigation framework. By recording the vector map offline, the framework initializes an optimal global route by giving any starting and ending position on the map. During runtime, the computer filters the real-time positioning data from DGPS and tracks the planned path according to geometric rules. Besides, when the Lidar/vision subsystem detects any obstacles, the method dynamically adjusts the local path to realize obstacle avoidance. Demonstrated on an AV testing platform, the proposed method calculates the angle of the steering wheel and transmits the actual commands through CAN bus interface.