Deep Multi-Agent Reinforcement Learning for Highway On-Ramp Merging in Mixed Traffic

On-ramp merging is a challenging task for autonomous vehicles (AVs), especially in mixed traffic where AVs coexist with human-driven vehicles (HDVs). In this paper, we formulate the mixed-traffic highway on-ramp merging problem as a multi-agent reinforcement learning (MARL) problem, where the AVs (on both merge lane and through lane) collaboratively learn a policy to adapt to HDVs to maximize the traffic throughput. We develop an efficient and scalable MARL framework that can be used in dynamic traffic where the communication topology could be time-varying. Parameter sharing and local rewards are exploited to foster inter-agent cooperation while achieving great scalability. An action masking scheme is employed to improve learning efficiency by filtering out invalid/unsafe actions at each step. In addition, a novel priority-based safety supervisor is developed to significantly reduce collision rate and greatly expedite the training process. A gym-like simulation environment is developed and open-sourced with three different levels of traffic densities. We exploit curriculum learning to efficiently learn harder tasks from trained models under simpler settings. Comprehensive experimental results show the proposed MARL framework consistently outperforms several state-of-the-art benchmarks.

I. Introduction

Autonomous vehicle (AV) technologies, such as Tesla Autopilot [1] and Baidu Apollo [2], have already been deployed in (semi-)autonomous vehicles on real-world roads. Despite the great advances over the past decade that have made this possible, the number of traffic accidents involving AVs are increasing in recent years [3], [4]. The accidents are often caused by the inability of AVs to timely react to the dynamic driving environment, especially in mixed traffic with both AVs and human-driven vehicles (HDVs); the AVs need not only to react to road objects but also to attend to the behaviors of HDVs. Among the many challenging driving scenarios, highway on-ramp merging is one of the most difficult tasks for AVs [5], [6], which is the topic of this paper.