Driving risk emerges from the required two-dimensional joint evasive acceleration

2026-04-20

AI Summary

This study addresses the limitations of existing autonomous driving safety assessment metrics—such as Time-to-Collision (TTC)—which oversimplify risk into a one-dimensional scalar and fail to accurately capture the dynamic, two-dimensional nature of real-world collision-avoidance scenarios. To overcome this, the work proposes “Evasive Acceleration” (EA) as a novel risk metric that quantifies the minimum constant relative acceleration vector magnitude required to render an interaction collision-free. EA establishes the first hyperparameter-free, physically interpretable framework for two-dimensional risk quantification. Validated across five public datasets and over 600 real-world crashes, EA consistently delivers the earliest statistically significant warnings across all thresholds, achieves superior collision discrimination performance, and improves information retention by 54.2%–241.4% compared to state-of-the-art methods.

Technology Category

• Multiagent Systems: Adversarial Agents
• Search and Optimization: Metareasoning and Metaheuristics
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Metrics for user behavior and evaluating success

Abstract

Most autonomous driving safety benchmarks use time-to-collision (TTC) to assess risk and guide safe behaviour. However, TTC-based methods treat risk as a one-dimensional closing problem, despite the inherently two-dimensional nature of collision avoidance, and therefore cannot faithfully capture risk or its evolution over time. Here, we report evasive acceleration (EA), a hyperparameter-free and physically interpretable two-dimensional paradigm for risk quantification. By evaluating all possible directions of collision avoidance, EA defines risk as the minimum magnitude of a constant relative acceleration vector required to alter the relative motion and make the interaction collision-free. Using interaction data from five open datasets and more than 600 real crashes, we derive percentile-based warning thresholds and show that EA provides the earliest statistically significant warning across all thresholds. Moreover, EA provides the best discrimination of eventual collision outcomes and improves information retention by 54.2-241.4% over all compared baselines. Adding EA to existing methods yields 17.5-95.5 times more information gain than adding existing methods to EA, indicating that EA captures much of the outcome-relevant information in existing methods while contributing substantial additional nonredundant information. Overall, EA better captures the structure of collision risk and provides a foundation for next-generation autonomous driving systems.

Problem

• driving risk
• collision avoidance
• time-to-collision
• two-dimensional risk
• autonomous driving safety

Innovation

• evasive acceleration
• two-dimensional risk
• collision avoidance
• autonomous driving safety
• information retention