Trailer-aware End-to-end Autonomous Driving for Tractor-Trailers with Deep Reinforcement Learning
Congfei Li, Yang Li, Peigen Liu, Rongqi Gu, Zuolei Sun, Yuxiang Sun
AI summary
Problem
Existing end-to-end autonomous driving methods focus on small vehicles and ignore the underactuated dynamics and extended wheelbase of tractor-trailers, causing trailer collisions and unsafe lane encroachments during tight turns.
Approach
The authors derive a trailer-aware reward function from planar rigid-body kinematics and train an end-to-end driving policy using Soft Actor-Critic reinforcement learning combined with imitation learning in a custom CARLA simulator.
Key results
- Designed a novel trailer-aware reward function for safe interaction navigation
- Investigated the impact of reference point placement on turning envelopes
- Built a custom QTruck model for CARLA and conducted closed-loop experiments
- Achieved 100% success rate and 84% no-crash rate, outperforming small-vehicle baselines
Why it matters
Provides a critical framework for safely deploying autonomous articulated trucks in complex urban environments where conventional driving policies fail.
Abstract
End-to-end autonomous driving has been greatly advanced in recent years. However, most of existing work focuses on small vehicles (e.g., cars). Driving articulated trucks, such as tractor-trailers, still remains less being explored. The underactuated nature and extended wheelbase of tractor- trailers pose considerable driving challenges, especially when navigating narrow roads. For example, when a left-hand-drive tractor-trailer makes a right turn on a two-way two-lane narrow road, the tractor usually needs to encroach some spaces in the opposing lane. Otherwise, the trailer may have insufficient spaces to turn right and strike curbside objects. To provide a solution to this problem, we employ deep reinforcement learning to train an end-to-end autonomous driving policy with a trailer-aware reward function. Through planar rigid- body kinematics analysis, we locate the reference points on the tractor and the trailer. We also build a tractor-trailer model for CARLA. Experimental results demonstrate the effectiveness and superiority of our method in CARLA.