Radar Accuracy Modeling and Its Application to Object Vehicle Tracking

Abstract

For advanced active safety systems and autonomous vehicles, it is essential to predict target vehicle behaviors. Radar is a reliable vehicular sensor, but it has a degree of measurement uncertainty because of its operational mechanism and wandering reflection point. To deal with this problem, this paper proposes a radar accuracy model for object vehicle tracking to reduce the tracking error of the radar system. First, we divide the area uniformly and identify the surrounding vehicles using information, including relative position and relative velocity. We then analyze the error characteristics of each uniformly divided area. Next, we develop an occupancy zone-based Kalman filter for each area based on the error characteristics. We solve the problem that one cannot assign an exact zone for the object vehicle due to radar error by using interacting multiple model (IMM) filter. The performance of the proposed method is validated via experiments using radar and ground truth data.