Map-aided GPS/INS localization using a low-order constrained Unscented Kalman Filter
Kang Li, Han-Shue Tan, J. Karl Hedrick · 2009
This paper presents a map-aided GPS/INS localization system developed for a map-based driver safety assistance system. A low-order constrained Unscented Kalman Filter (CUKF) is adopted to fuse the GPS, INS and high-accuracy map data to provide robust lane-level vehicle position and heading estimates for time-critical safety applications. The road data from digital map is formulated as state constraints including the heading and lateral in-lane/road position constraints. Compared to the other nonlinear state estimation approaches with or without constraints, such as the Particle Filter (PF), Moving Horizon Estimation (MHE) and Extended Kalman Filter (EKF), the proposed low-order constrained Unscented Kalman Filter has the merits of improved computational tractability, higher accuracy, ease of implementation and better state constraint handling capability. Testing results using real sensor data are compared with a previously developed EKF-based GPS/INS positioning system to demonstrate the effectiveness of the new CUKF approach.