Vehicle position estimation using nonlinear tire model for autonomous vehicle

To achieve technical perfection of the Advanced driver assistant system (ADAS), accurate analysis of the vehicle's position is essential. For this, conventionally, sensor fusion has been carried out using a general GPS and general Inertial measurement unit (IMU), but the position accuracy decreases because of inertial sensor accumulation. Furthermore, because a vehicle tire model is analyzed by linearization and using a bicycle model, the position error increases. To solve this, in this study, a fusion algorithm was proposed by using an extended Kalman filter based on the non-linear tire model for the vehicle state information and by using the general GPS position information provided by the electric stability program of the vehicle. The fusion algorithm proposed in this study allowed us to suggest a position error correction method corresponding to a high precision Differential global positioning system (DGPS) within 1 m.