Developing a safe and highly effective policy for autonomous vehicles (AVs) continues to pose a significant challenge in machine learning. In this study, we propose a novel reinforcement learning approach with reward machine. By tailoring the reward function to the specific needs of AVs in highway scenarios, we enable them to make more informed and efficient decisions. Our focus is on designing a reward function to formalize traffic rules, which is crucial for achieving safe and effective AV behavior on highways. To address this problem, we propose several innovative ideas that go beyond existing algorithmic techniques, specifically aimed at facilitating exploration and exploitation on different operations. To our knowledge, this is the first reinforcement learning algorithm that can integrate the safe distance with autonomous highway driving, aiming at the Vienna Convention on road traffic. Experimental results demonstrate the effectiveness of the proposed approach, which significantly improves AVs’ safety and performance on highways.
Reward Machine Reinforcement Learning for Autonomous Highway Driving: An Unified Framework for Safety and Performance
27.10.2023
3256676 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
DEEP REINFORCEMENT LEARNING WITH ENHANCED SAFETY FOR AUTONOMOUS HIGHWAY DRIVING
British Library Conference Proceedings | 2020
|ROAD: Responsibility-Oriented Reward Design for Reinforcement Learning in Autonomous Driving
ArXiv | 2025
|