Intersection collision prediction and prevention based on vehicle-to-vehicle (V2V) and cloud computing communication

Area of Science:

• Intelligent Transportation Systems
• Traffic Safety Engineering
• Machine Learning for Transportation

Background:

• Rising vehicle complexity necessitates advanced traffic safety management.
• Effective collision risk assessment is crucial for road efficiency and safety.
• Current systems face challenges in predicting complex traffic scenarios.

Purpose of the Study:

• To develop an intelligent traffic collision prediction model.
• To leverage vehicle-to-vehicle (V2V) communication and graph attention networks (GAT) for risk assessment.
• To enhance decision-making capabilities in intelligent transportation systems.

Main Methods:

• Utilizing V2V communication to gather vehicle data (trajectory, speed, acceleration, relative position).
• Constructing a graph representation of the traffic environment.
• Applying GAT to extract inter-vehicle interaction features.
• Optimizing driving strategies using deep reinforcement learning (DRL).

Main Results:

• The proposed framework achieves over 80% collision recognition accuracy (true warning rate).
• Analysis demonstrates the model's efficacy and robustness on public and real-world datasets.
• False detection metrics were thoroughly evaluated.

Conclusions:

• The novel approach significantly enhances traffic safety in intelligent transportation systems.
• The model improves decision-making efficiency for traffic management.
• This research offers a robust technological advancement for collision prediction.