An Enhanced Graph Representation for Machine Learning Based Automatic Intersection Management

TL;DR

This paper introduces an improved graph-based scene representation combined with reinforcement learning for automatic intersection management, leading to reduced delays and good generalization across different intersection layouts.

Contribution

It enhances previous graph representations with edge features and updates the neural network architecture, advancing learning-based intersection management methods.

Findings

Significant delay reduction compared to traditional methods.

Effective generalization to unseen smaller intersection layouts.

Maintains performance within certain limits on larger intersections.

Abstract

The improvement of traffic efficiency at urban intersections receives strong research interest in the field of automated intersection management. So far, mostly non-learning algorithms like reservation or optimization-based ones were proposed to solve the underlying multi-agent planning problem. At the same time, automated driving functions for a single ego vehicle are increasingly implemented using machine learning methods. In this work, we build upon a previously presented graph-based scene representation and graph neural network to approach the problem using reinforcement learning. The scene representation is improved in key aspects by using edge features in addition to the existing node features for the vehicles. This leads to an increased representation quality that is leveraged by an updated network architecture. The paper provides an in-depth evaluation of the proposed method against baselines that are commonly used in automatic intersection management. Compared to a traditional signalized intersection and an enhanced first-in-first-out scheme, a significant reduction of induced delay is observed at varying traffic densities. Finally, the generalization capability of the graph-based representation is evaluated by testing the policy on intersection layouts not seen during training. The model generalizes virtually without restrictions to smaller intersection layouts and within certain limits to larger ones.