System for following a vehicle ahead utilizing onboard sensors for multi-vehicle transportation

At vehicle transportation sites, each vehicle requires manual handling by workers; however, severe labor shortages have been reported. In closed environments such as finished vehicle yards, labor savings can be expected by implementing a system for following a vehicle ahead automatically that enables multiple vehicles to be transported simultaneously by one human-driven leading vehicle. Although the preceding vehicle following control has been researched, these studies focus on low-speed scenarios and have noted a tendency for increasing trajectory deviations from the vehicle ahead. The aim of this study was to develop the system for following a vehicle ahead under various speed and curvature conditions. We proposed the algorithm for estimating the preceding vehicle’s trajectory based on tail position data obtained via onboard sensors. The point cloud of the preceding vehicle’s trajectory was obtained by correcting past tail positions using the ego vehicle’s displacement calculated from a vehicle model and applying coordinate transformations. To calculate the target curvature for steering control, the point cloud was approximated using a cubic spline curve. To verify the effectiveness, simulations and real-vehicle experiments were conducted in an environment consisting of one leading vehicle and five following vehicles. In the 90° left-turn task, both experiments achieved a trajectory deviation of less than 0.35 m from the fifth following vehicle to the leading vehicle. Additionally, the validity of the proposed method was confirmed by real-vehicle trials conducted under various speed and curvature conditions.