Vehicle position estimation using tire model

Abstract

GPS is being widely used in the location estimation technology, which is essential for stable driving of autonomous vehicle. However, GPS has problems such as reduction in location accuracy during abrupt vehicle behavior at high speed, and limitations such as signal interruption in tunnels and downtown areas. To overcome this problem, an algorithm that combines various sensor information and longitudinal/lateral slip is required. This paper proposes a three-degree of freedom (3-DoF) vehicle dynamics model, in which Dugoff’s tire model is applied, and an algorithm, which combines various sensor information inside the vehicle by using extended Kalman filter. The performance of proposed location estimation algorithm was analyzed and evaluated through simulations. As a result, it is confirmed that the location estimation result of proposed algorithm is more accurate than that of method using GPS even during abrupt changes in motion.