ML Predictive Framework for Signalized Intersection Safety
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing traffic management systems rely heavily on historical crash data, which is often limited, outdated, or unavailable, making it difficult to accurately predict and prevent collisions at signalized intersections.
Innovation Solution
A modeling framework that utilizes connected vehicle event data, speed data, and other relevant information to develop machine learning-based predictive models, which can identify critical signalized intersections and predict traffic safety risks even in the absence of historical crash data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If historical crash data is used to identify safety hotspots, then crash analysis can be performed, but the analysis is reactive and includes time lag
Solution Approach 1:
The system performs preliminary safety evaluations by analyzing connected vehicle event data, speed data, and intersection characteristics before crashes occur. Machine learning models predict potential crash risks at signalized intersections based on real-time and historical data patterns, enabling proactive identification of safety hotspots rather than waiting for crashes to be reported and processed.
2Reliability
If historical crash data is used to identify safety hotspots, then crash statistics can be analyzed, but minor crashes and near misses are not reflected
Solution Approach 1:
The system introduces connected vehicle event data and speed data as intermediary data sources that capture safety-relevant information before official crash reports are generated. These data types include hard braking, hard acceleration, and speeding events that can indicate potential crashes or near misses, providing a more complete picture of intersection safety risks without relying solely on post-crash reporting.
3Measurement precision
If traditional crash data collection methods are used, then official crash records can be obtained, but data collection is incomplete and has time lag
Solution Approach 1:
The system replaces the traditional mechanical data collection process (relying on police reports and official crash records) with electronic data collection from connected vehicles. Onboard sensors continuously capture event data, speed information, and location data, which are transmitted and analyzed in real-time or near real-time, eliminating the months-long time lag associated with traditional crash data collection and verification processes.
4Reliability
If machine learning models are trained on historical crash data, then predictive accuracy can be improved, but the models cannot be applied where crash data is unavailable
Solution Approach 1:
The machine learning models are designed to be universally applicable across different locations and time periods. The models can process multiple data types including connected vehicle event data, speed data, intersection geometry, and traffic volume data, making them adaptable to various contexts. This multi-functionality allows the same model framework to be applied both where historical crash data exists and where only alternative data sources are available.
Data Source
AI summary
A machine learning-based modeling framework provides an approach for quickly identifying and prioritizing intersection safety improvements, and for analyzing adverse traffic events such as collisions and crashes at signalized intersections within a transportation network, to generate risk scores for intersection safety performance screening in a large such network. The framework enables machine learning models to be run using current data to identify dangerous locations and treat them with traffic safety measures, without having available up-to-date or recent crash data. These machine learning models also account for dangerous behavior by considering speed and excessive hard braking and hard acceleration to produce more accurate risk scores.