ROITP: Road Obstacle-Involved Trajectory Planner for Autonomous Trucks

Abstract Autonomous trucks have the potential to enhance both safety and convenience in intelligent transportation. However, their maximum speed and ability to navigate a variety of driving conditions, particularly uneven roads, are limited by a high center of gravity, which increases the risk of rollover. Road bulges, sinkholes, and unexpected debris all present additional challenges for autonomous trucks' operational design, which current perception and decision-making algorithms often overlook. To mitigate rollover risks and improve adaptability to damaged roads, this paper presents a novel Road Obstacle-Involved Trajectory Planner (ROITP). The planner categorizes road obstacles using a learning-based algorithm. A discrete optimization algorithm selects a multi-objective optimal trajectory while taking into account constraints and objective functions derived from truck dynamics. Validation across various scenarios on a hardware-in-loop platform demonstrates that the proposed planner is effective and feasible for real-time implementation.