Roundabout Dilemma Zone Data Mining and Forecasting with Trajectory Prediction and Graph Neural Networks

TL;DR

Problem: Roundabout traversal creates dilemma zones that jeopardize safety for autonomous and human drivers. Approach: a data-mining pipeline built on a graph-structured trajectory predictor ST++ that fuses agent dynamics and semantic maps to forecast DZ events and detect abnormal driving. Contributions: (1) DZ boundary estimation with virtual yellow-light labeling, (2) an abnormal-driving detector based on trajectory deviations, (3) a graph neural network for DZ forecasting producing $P_{Dilemma}$ and $P_{Causal}$ at each moment. Findings: on real-world CAROM Air drone data, the method achieves high precision with a low false-positive rate around 0.1 and yields improved trajectory accuracy when using DZ features. Significance: enables proactive safety decisions in mixed traffic and informs ITS research and deployment for safer intersections.

Abstract

Traffic roundabouts, as complex and critical road scenarios, pose significant safety challenges for autonomous vehicles. In particular, the encounter of a vehicle with a dilemma zone (DZ) at a roundabout intersection is a pivotal concern. This paper presents an automated system that leverages trajectory forecasting to predict DZ events, specifically at traffic roundabouts. Our system aims to enhance safety standards in both autonomous and manual transportation. The core of our approach is a modular, graph-structured recurrent model that forecasts the trajectories of diverse agents, taking into account agent dynamics and integrating heterogeneous data, such as semantic maps. This model, based on graph neural networks, aids in predicting DZ events and enhances traffic management decision-making. We evaluated our system using a real-world dataset of traffic roundabout intersections. Our experimental results demonstrate that our dilemma forecasting system achieves a high precision with a low false positive rate of 0.1. This research represents an advancement in roundabout DZ data mining and forecasting, contributing to the assurance of intersection safety in the era of autonomous vehicles.

Figures (7)

• Figure 1: An illustration of a roundabout dilemma zone event. Here, a car in the yellow zone is insinuating a "moving yellow light", creating a dilemma zone that influences the car in the red zone entering the roundabout.
• Figure 2: An example of the driving behavior in a roundabout DZ: (1) Vehicle 'B' inside the roundabout triggers a DZ as the area circled in red. (2) Vehicle 'A' enters the DZ, and its trajectory is predicted. (3) Vehicle 'A' rapidly decelerates, causing extensive deviation between the predicted trajectory and the actual trajectory. (4) Upon exit, vehicle 'A' resumes, and its predicted trajectory agrees with the actual trajectory.
• Figure 3: Formulation of the roundabout DZ.
• Figure 4: Dilemma forecasting with graph neural network.
• Figure 5: Abnormal driving events detected using trajectory deviation from the prediction with dilemma (red) and without dilemma (yellow).