Attitude Estimation for Vehicles With Partial Inertial Measurement

In this paper, a novel method for the estimation of the attitude of an automotive vehicle is presented. The algorithm uses low-cost sensors, namely, a Global Positioning System (GPS) receiver and a three-axis accelerometer. It employs a kinematic model of the vehicle, which is augmented by unknown parameters of the system. An extended Kalman filter (EKF) is employed, which produces estimates of the vehicle attitude, as well as the installation angles of the sensor unit with respect to the vehicle. Compared with existing approaches, it does not require knowledge of the road tilt angle, and it automatically compensates for different installation angles. It is therefore particularly well suited for applications where there is no control over how the unit is installed in the vehicles, as is the case with personal navigation assistants. The method is numerically robust by using nonsingular attitude parametrization; results are given for simulations and real-world experiments.