Curve modeled lane and stop line detection based GPS error estimation filter

Street and angle information can be obtained from vision sensor and center of lane. Lane following system can be easily implemented through the use of controller making these two estimates 0. However, this cannot be applied into intersection without lane. Therefore, this paper proposes a method of estimating GPS error through filter for enhanced performance of waypoint following Guidance methods. In a curved section, straight line detection method is not desirable. In this method, errors occur in lateral distance estimation due to model mismatch. Therefore, this paper evaluates performance of proposed GPS error estimation filter based on lane and stop line detection of curved lane models in order to enhance performance of waypoint following and applying it to autonomous vehicles in experimental regions.