Enhancing Intelligent Vehicle Lane-Change Strategies Based on Deep Reinforcement Learning

Ruihui Li

2024

Abstract

The evolution of intelligent vehicle lane-change strategies, propelled by advancements in automated driving technology, underscores the significance of efficient road utilization, traffic congestion reduction, and driving safety. This paper investigates the impact of varying penalty values for unnecessary lane changes at different speed limits on lane-change decisions, aiming to ascertain effective strategies. Employing deep reinforcement learning, this study simulates and analyses vehicle lane-change behaviours. Initially, a simulated traffic environment is constructed, and a reward system is defined to reflect different speed limits and unnecessary lane-change penalties. Utilizing the deep deterministic policy gradient (DDPG) algorithm, vehicles are trained to optimize lane-change strategies across diverse scenarios. Evaluation based on average rewards demonstrates that increasing the penalty for unnecessary lane changes enhances vehicle speed and facilitates safer time headway maintenance at both low and high-speed limits. Experimental findings indicate that adjusting the penalty effectively guides vehicles towards cautious lane-change decisions, thereby enhancing driving efficiency and safety. This discovery presents a novel adjustment mechanism for autonomous driving system decision algorithms and offers insights for the development of more intelligent traffic management systems, promoting enhanced road utilization alongside driving safety.

Paper Citation

in Harvard Style

Li R. (2024). Enhancing Intelligent Vehicle Lane-Change Strategies Based on Deep Reinforcement Learning. In Proceedings of the 1st International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI; ISBN 978-989-758-713-9, SciTePress, pages 200-207. DOI: 10.5220/0012923000004508

in Bibtex Style

@conference{emiti24,
author={Ruihui Li},
title={Enhancing Intelligent Vehicle Lane-Change Strategies Based on Deep Reinforcement Learning},
booktitle={Proceedings of the 1st International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI},
year={2024},
pages={200-207},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012923000004508},
isbn={978-989-758-713-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI
TI - Enhancing Intelligent Vehicle Lane-Change Strategies Based on Deep Reinforcement Learning
SN - 978-989-758-713-9
AU - Li R.
PY - 2024
SP - 200
EP - 207
DO - 10.5220/0012923000004508
PB - SciTePress