Ma, Pengtao (DeepWay), Sun, Lei (DeepWay), Wei, Shouyang (Beijing DeepWay Corporation), Wan, Diana (University of British Columbia), Ding, Feng (Deepway Corporation), Zhang, Donghao (Deepway), Tian, Shan (Deepway)

Real-Time Optimization-Based Path Planning for Autonomous Semi-Trailer Trucks

Scheduled for presentation during the Regular Session "Control of heavy vehicles" (FrAT4), Friday, September 27, 2024, 11:10−11:30, Salon 7

2024 IEEE 27th International Conference on Intelligent Transportation Systems (ITSC), September 24- 27, 2024, Edmonton, Canada

This information is tentative and subject to change. Compiled on December 26, 2024

Abstract

Autonomous semi-trailer trucks have great potential to offer safer and more efficient transportation. Path planning is an important part of autonomous driving, which aims to generate an optimal path for vehicles to avoid collision risks and keep them as centered in the lane as possible. However, achieving an optimal path for semi-trailer trucks is challenging due to the complex kinematics, the large vehicle dimensions and the trade-off between model complexity and real-time capability. In this work, we propose a novel real-time optimization-based path planning method to address these problems. This approach involves modeling the entire tractor-trailer system with positioning of all axles and corners. This detailed model enables more accurate path planning, allowing for full utilization of the drivable space while satisfying all physical constraints. The modeling is approximated and simplified by assuming equal curvature and using the law of cosines, which greatly reduces computation burden with slightly sacrificing modeling accuracy. Then we construct an optimization problem with strict collision avoidance constraints and soft lane centering preferences. This allows the truck's wheels to temporarily exceed the lane boundaries in certain scenarios like tight bends or narrow roads, improving its passing ability. The optimization problem is solved using high-efficient Augmented Lagrange Multiplier method. We demonstrate the performance of the proposed method with simulations and real semi-trailer truck experiments. The results show that the proposed method is efficient and accurate for real-time application. It can significantly improve the vehicle behavior in terms of obstacle avoidance and lane centering.