Robust vehicle localization with GPS dropouts

Abstract

Vehicle navigation systems rely on the correct positioning information and GPS is the best solution. However, in situations of GPS outage, like in tunnels and dense woods, the vehicle can lose track quickly. Modern vehicle navigation systems are mostly based on the vehicle dynamics; proprietary algorithms keep navigating for quite some time. But these solutions are expensive and add much to a vehicle's cost. This paper proposes a Kalman filter based dead-reckoning algorithm that fuses GPS information with the orientation information from a cheap IMU/INS, and the vehicle's speed accessed from its ECU. This low-cost system uses GPS and IMU/INS in a loosely coupled manner along with vehicle's speed and keeps supplying a quite accurate position information with GPS outage for significantly long intervals. With proper tuning and initialization, the proposed scheme has a potential to keep working in a completely GPS-denied situation.