An Observation Model Based on Polyline Map for Autonomous Vehicle Localization

Solution of the localization problem for autonomous vehicle navigation is an urgent requirement. In the wake of this requirement a new map-based method for the localization of autonomous vehicles using the extended Kalman Alter (EKF) is proposed. Formulation of the EKF equations is based upon a 4-wheel vehicle equipped with encoders, laser rangefinder and a polyline map. The observation model is comprised of special scanned points. The equations are derived for both range and bearing to form an effective observation model for the EKF estimator. Once the matching is set up, the pose predicted by dead reckoning will be well corrected for a robust localization.