Research on a Driver's Perceived Risk Prediction Model Considering Traffic Scene Interaction

2025-03-06

AI Summary

To address inaccurate driver-perceived risk prediction in conditional automated driving, this paper proposes a human–machine collaborative risk prediction model. Methodologically, it introduces the first systematic decoupling of subjective factors (e.g., driver-specific traits) from objective factors (e.g., ego-vehicle dynamics, traffic environment, and multi-agent interactions). It further designs a personalized modeling strategy coupled with an interaction-aware deep temporal network, enabling effective fusion of heterogeneous multi-source features and incorporating a parameter-adaptive mechanism. Evaluated on standard benchmark datasets, the proposed model achieves a 10.0% improvement in prediction accuracy over state-of-the-art methods. This advancement significantly enhances the timeliness of risk anticipation and overall system safety, thereby providing critical support for trustworthy human–machine cooperative decision-making.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Reasoning under Uncertainty: Relational Probabilistic Models
• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

In the field of conditional autonomous driving technology, driver perceived risk prediction plays a crucial role in reducing traffic risks and ensuring passenger safety. This study introduces an innovative perceived risk prediction model for human-machine interaction in intelligent driving systems. The model aims to enhance prediction accuracy and, thereby, ensure passenger safety. Through a comprehensive analysis of risk impact mechanisms, we identify three key categories of factors, both subjective and objective, influencing perceived risk: driver's personal characteristics, ego-vehicle motion, and surrounding environment characteristics. We then propose a deep-learning-based risk prediction network that uses the first two categories of factors as inputs. The network captures the interactive relationships among traffic participants in dynamic driving scenarios. Additionally, we design a personalized modeling strategy that incorporates driver-specific traits to improve prediction accuracy. To ensure high-quality training data, we conducted a rigorous video rating experiment. Experimental results show that the proposed network achieves a 10.0% performance improvement over state-of-the-art methods. These findings suggest that the proposed network has significant potential to enhance the safety of conditional autonomous driving systems.

Problem

• Predict driver's perceived risk in autonomous driving scenarios.
• Enhance risk prediction accuracy using deep-learning techniques.
• Improve safety by modeling driver-specific traits and traffic interactions.

Innovation

• Deep-learning-based risk prediction network
• Personalized modeling with driver-specific traits
• Interactive traffic participant relationship analysis