Connected multi-vehicle crash risk assessment considering probability and intensity

Area of Science:

• Intelligent Transportation Systems
• Traffic Safety Engineering
• Vehicle Dynamics

Background:

• Accurate driving risk assessment is crucial for vehicle collision avoidance and traffic safety.
• Current methods often focus only on pre-crash indicators (e.g., time-to-collision) and neglect crash severity.
• A comprehensive approach integrating both pre- and post-crash factors is needed for robust risk evaluation.

Purpose of the Study:

• To propose a novel collision risk assessment model for vehicle-to-vehicle communication environments.
• To achieve a more scientific driving risk assessment by integrating collision probability (pre-crash) and intensity (post-crash).
• To enhance trajectory prediction accuracy by considering driving intentions and vehicle interactions.

Main Methods:

• Developed a trajectory prediction model incorporating driving intentions and utilizing a social tensor pool for vehicle interactions.
• Assessed collision likelihood by analyzing conflicts between predicted and candidate vehicle trajectories.
• Determined collision intensity based on vehicle driving states.
• Validated the model using publicly available unmanned aerial vehicle (UAV)-based traffic data.

Main Results:

• The proposed trajectory prediction model achieved low prediction errors: 0.68 m Root Mean Square Error (RMSE) and 1.34 Negative Log-Likelihood (NLL) for 3-second trajectories.
• The integrated risk assessment model demonstrated superior performance compared to existing models.
• Quantitative results confirmed the model's ability to scientifically assess vehicle travel risk.

Conclusions:

• The developed model provides a more scientific and comprehensive approach to collision risk assessment.
• Integrating pre-crash probability and post-crash intensity significantly improves the accuracy and relevance of driving risk evaluation.
• The model holds promise for enhancing traffic safety and enabling effective collision avoidance systems.