Adaptive level lane estimation policy of automotive systems moving within arbitrary road networks
Résumé
The problem of a Vehicle Positioning System (VPS) is adressed. A specifically developed approach based on the cause-and-effect principle defines a closed-loop policy determining the lane level location of a sensor equipped vehicle moving within arbitrary road networks. After exploration of processed data provided by a smart camera and a laser detector appropriate information is routed to the decision making scheme. Based on identification of particular mobile objects circulating within the current and/or opposite flow the vehicle lane level location is determined when the car circulates within highway stretches or urban areas. Contrary to other schemes no additional investment on costly devices and complementary tools is required. The reduced computational complexity makes the deployment of the suggested methodology efficient for embedded technologies and can be utilized independently or as a complementary tool. Major restrictions of previous works of the same authors requiring divided freeway infrastructures are now overcome. A first appraisal of the system effectiveness is achieved though realistic simulation data fed by complex real recordings. Multiple drawbacks involved with information collected by professional driver behavior are minimized. Three case study scenarios, associated with different traffic and light intensity, resulting to a varying sensor behavioral functioning are considered. Adequate metrics appraise the efficiency of the suggested methodology.
Origine | Fichiers produits par l'(les) auteur(s) |
---|