Improving Urban Vehicle Localization with Traffic Sign Recognition

Accurate and reliable vehicle localization is an important requirement for many vehicular applications. In challenging environments like urban areas, the GNSS accuracy often degrades due to blocked or reflected satellite signals. To improve the standalone positioning accuracy, we propose a landmark based localization method using traffic signs. The traffic sign detections from a camera are associated with the landmarks in a pre-generated traffic sign database and correct the GNSS position of the vehicle within a Bayes Filter. This approach was evaluated with real-world data collected during a test drive in difficult urban scenarios and the results were compared with low-cost GNSS positioning.