Robust Multirate On-Road Vehicle Localization for Autonomous Highway Driving Vehicles

An on-road vehicle localization scheme is proposed to keep a track of an ego vehicle with respect to the ego/target lane center, using the results of the camera-based lane recognition. A vehicle's lateral motion model is first developed in terms of lane curve tracking error states, which is assumed to satisfy the clothoid road constraint. The proposed vehicle localization scheme computes vehicle's state, such that advanced state space control can be applied at a constant fast rate. Utilizing multirate characteristics of each sensor, a multirate decentralized state estimator is developed, which exhibits robust convergence against uncertain slow update period of the camera-based lane recognition. Performance and robustness of the proposed on-road vehicle localization scheme is verified through the computational and experimental tests using camera-based lane recognition.