An Approach for Navigating Autonomous Trucks in Unpaved Environments

Autonomous navigation in off-road environments presents unique challenges, including uneven terrain, unpredictable obstacles, and localization inaccuracies. This paper introduces a robust framework for autonomous truck navigation tailored to such scenarios. The system integrates a Model Predictive Control (MPC) method that enhances path-following precision and adaptability compared to conventional controllers like the Stanley Controller. The architecture combines a local planner and an MPC-based controller within an event-driven framework to ensure dynamic path planning, obstacle avoidance, and responsiveness to environmental uncertainties. The system was evaluated in a co-simulated environment using AVL VSM <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$^{\text {TM }}$</tex> for vehicle dynamics and CARLA for off-road terrain simulation. Results show that the MPC reduces cross-track errors and improves trajectory stability under varying conditions. Additionally, the framework demonstrated effective obstacle avoidance and adaptability to localization noise, maintaining safe and reliable operation.