AI2-Active Safety: AI-enabled Interaction-aware Active Safety Analysis with Vehicle Dynamics

TL;DR

The work addresses the need for proactive, uncertainty-aware active safety in complex highway environments. It introduces a two-module framework that couples a gradient-aware augmented bicycle model with a hypergraph Transformer to forecast multimodal ambient trajectories and compute stochastic, interaction-aware high-fidelity Time-to-Collision ($HF\text{-}TTC$) metrics via a Runge–Kutta-based integration of host dynamics. Key contributions include: (i) high-fidelity host dynamics with road grade, (ii) a hypergraph-based multi-vehicle predictor capturing group interactions, (iii) probabilistic trajectory decoding with mode probabilities, and (iv) stochastic $HF\text{-}TTC$ distributions for early, robust collision warnings. Experiments on NGSIM and HighD show improved prediction accuracy and earlier risk signaling compared to traditional TTC and pairwise graph models, highlighting the framework’s potential for safer, proactive decision-making in connected and autonomous traffic. The results underscore the importance of modeling group-wise interactions and uncertainty to produce reliable surrogate safety measures in real-world driving scenarios.

Abstract

This paper introduces an AI-enabled, interaction-aware active safety analysis framework that accounts for groupwise vehicle interactions. Specifically, the framework employs a bicycle model-augmented with road gradient considerations-to accurately capture vehicle dynamics. In parallel, a hypergraph-based AI model is developed to predict probabilistic trajectories of ambient traffic. By integrating these two components, the framework derives vehicle intra-spacing over a 3D road surface as the solution of a stochastic ordinary differential equation, yielding high-fidelity surrogate safety measures such as time-to-collision (TTC). To demonstrate its effectiveness, the framework is analyzed using stochastic numerical methods comprising 4th-order Runge-Kutta integration and AI inference, generating probability-weighted high-fidelity TTC (HF-TTC) distributions that reflect complex multi-agent maneuvers and behavioral uncertainties. Evaluated with HF-TTC against traditional constant-velocity TTC and non-interaction-aware approaches on highway datasets, the proposed framework offers a systematic methodology for active safety analysis with enhanced potential for improving safety perception in complex traffic environments.

Figures (9)

• Figure 1: Framework Architecture.
• Figure 2: Hypergraph Transformer Network.
• Figure 3: Graph and hypergraph.
• Figure 4: Hypergraph Relational Transformer.
• Figure 5: Stochastic High-fidelity Time-to-Collision.

Theorems & Definitions (1)

• Remark