Estimation of Trailer-Vehicle Articulation Angle Using 2D Point-Cloud Data

In the quest to achieving more autonomous features on articulated vehicles, such as backing up and trailer stability, the vehicles need to keep track of the angle of articulation for proper control. While there are existing approaches which estimate the angle, most of the methods require a sticker on the trailer or additional hardware, which adds to the many-sensors already on the vehicle. This study considers a typical single-trailer articulation. Our approach takes advantage of the existing radar sensors at the rear side of the vehicle. Fortunately, most vehicle manufacturers follow this design. This prevents sourcing additional hardware for the task. We demonstrate the estimation approach with simulated point clouds for one of the radars at the rear of the vehicle with a view to implementing the approach on other sensors. The methods used are ordinary least squares (OLS), principal component analysis (PCA), and maximum likelihood estimation (MLE). Based on the simulation results, all three methods were comparable for point clouds having a low variance along the vehicle's longitudinal axis. When the variance along the axis was increased, OLS had a reduced performance while the PCA and the MLE methods retained their comparable performance. Meanwhile, the MLE method required more computational resource than the PCA method.