Constrained unscented Kalman filter based fusion of GPS/INS/digital map for vehicle localization

Abstract

Accurate vehicle localization is very important for various applications of intelligent transportation systems (ITS) including cooperative driving, collision avoidance, and vehicle navigation. In this paper, a constrained unscented Kalman filter (CUKF) algorithm is proposed to fuse differential global position system (DGPS), inertial navigation system (INS) and digital map to estimate the vehicle states. Using the road geometry information obtained from a digital map database, some state constraints can be formed. The measurements of DGPS and INS are used to set up the dynamic and measurement equations of the nonlinear filtering. The vehicle states are first estimated by the loosely coupled DGPS/INS system and the unconstrained UKF, and then the UUKF estimates are projected into the state constraints to obtain the final CUKF estimates. Synthetic and real data are used to evaluate the performance of the CUKF algorithm for fusing DGPS, INS and digital map.