Rectangular Geometric Constraints Based Vehicle Tracking Method for Automotive Radar

State-of-the-art extended object trackers (EOT) can jointly estimate both the target dynamics and the shape using high-resolution radar point clouds. This work addresses the automotive radar target tracking problem and proposes the EOT approach, considering a strictly rectangular vehicle shape. The proposed approach directly estimates the rectangular shape length and width and, therefore, is computationally efficient. Unlike conventional EOT approaches that consider elliptical or star-convex shapes, the proposed innovative measurement model leverages a customized radial function enforcing rectangular geometric constraints (RGC) that ensure the strictly rectangular shape. It is proposed that the strong nonlinearity of the RGCbased measurement model can be addressed using the unscented Kalman filter (UKF) for efficient tracking of rectangular objects. The performance of the proposed approach is evaluated via simulations. It shows that the proposed RGC-UKF approach outperforms other state-of-the-art approaches in estimating both the centroid and rectangular shape.