Enhancing High-Speed Cruising Performance of Autonomous Vehicles through Integrated Deep Reinforcement Learning Framework

TL;DR

This work tackles safe, high-speed autonomous driving in mixed traffic by proposing an integrated framework that fuses behavioral decision-making, IRL-informed path planning, and MPC-based motion control. The decision module uses a Bootstrapped DQN to decide between lane-keeping and lane-changing while generating motion-control weights, and the path-planning module learns human-like lane-change preferences via IRL represented by reward weights $\varpi$. The motion-control module employs MPC to track planned trajectories under vehicle dynamics and safety constraints, ensuring feasible and stable operation. Results show that the integrated approach yields higher cruising performance and rewards than a sequential baseline, and Bootstrapped DQN provides superior exploration and lower collision rates, illustrating practical benefits for safe, efficient mixed-traffic AV deployment.

Abstract

High-speed cruising scenarios with mixed traffic greatly challenge the road safety of autonomous vehicles (AVs). Unlike existing works that only look at fundamental modules in isolation, this work enhances AV safety in mixed-traffic high-speed cruising scenarios by proposing an integrated framework that synthesizes three fundamental modules, i.e., behavioral decision-making, path-planning, and motion-control modules. Considering that the integrated framework would increase the system complexity, a bootstrapped deep Q-Network (DQN) is employed to enhance the deep exploration of the reinforcement learning method and achieve adaptive decision making of AVs. Moreover, to make AV behavior understandable by surrounding HDVs to prevent unexpected operations caused by misinterpretations, we derive an inverse reinforcement learning (IRL) approach to learn the reward function of skilled drivers for the path planning of lane-changing maneuvers. Such a design enables AVs to achieve a human-like tradeoff between multi-performance requirements. Simulations demonstrate that the proposed integrated framework can guide AVs to take safe actions while guaranteeing high-speed cruising performance.

Figures (15)

• Figure 1: Vehicle dynamics model.
• Figure 2: The Systematic framework of the proposed method.
• Figure 3: The network structure of the Bootstrapped DQN.
• Figure 4: The candidate paths of path-planning scenario.
• Figure 5: Driving simulator platform.