Vehicle Collision Prediction under Reduced Visibility Conditions

Area of Science:

• Intelligent Transportation Systems (ITS)
• Traffic Safety Engineering
• Environmental Perception for Autonomous Systems

Background:

• Radar-based collision warning systems in ITS are significantly degraded by adverse weather conditions like fog and heavy rain.
• Existing Vehicle-to-Everything (V2X) communication methods for rear-end collision avoidance still require improvement, particularly concerning weather impacts on human response times.
• Low visibility conditions pose a critical challenge for accurate and timely collision detection and warning.

Purpose of the Study:

• To design and evaluate a Visibility-based Collision Warning System (ViCoWS) for enhanced rear-end collision detection under adverse weather.
• To develop predictive models for velocity, headway distance, and prediction horizon estimation that adapt to varying visibility.
• To provide real-time, weather-adaptive collision avoidance warnings to drivers.

Main Methods:

• Proposed a ViCoWS design integrating four key models: prediction horizon estimation, velocity prediction, headway distance prediction, and rear-end collision warning.
• Utilized historical velocity data to predict future velocity trends.
• Estimated the prediction horizon dynamically based on current weather and visibility conditions.

Main Results:

• The velocity prediction model achieved a mean absolute percentage error of less than 11%.
• In heavy fog (120m visibility), ViCoWS provided warnings up to 4.5 seconds before a potential collision, significantly outperforming the FCPI method (0.6s).
• In medium fog (160m visibility), ViCoWS offered 2.1s warnings, while FCPI provided only 0.6s. For congested traffic, ViCoWS gave 1.9s warnings compared to FCPI's 1.2s.

Conclusions:

• ViCoWS effectively enhances collision detection and warning capabilities in low visibility conditions by adapting to real-time weather changes.
• The system offers substantially earlier warnings compared to conventional methods like FCPI, improving driver reaction time and road safety.
• ViCoWS demonstrates a promising approach for intelligent transportation systems to mitigate risks associated with rear-end collisions in adverse weather.