Research on Pedestrian Crossing Decision Models and Predictions Based on Machine Learning.

Area of Science:

• Intelligent Transportation Systems
• Smart City Technology
• Artificial Intelligence in Traffic Management

Background:

• Road traffic safety is critical in smart cities, necessitating advanced solutions.
• Pedestrian crossing behavior analysis is vital for intelligent transportation systems.
• Traditional models struggle with the complexity of pedestrian crossing dynamics.

Purpose of the Study:

• To optimize pedestrian crossing predictions using machine learning (ML).
• To enhance the accuracy of judging and simulating pedestrian violations.
• To apply ML models for improved traffic simulation and safety.

Main Methods:

• Utilized OpenCV for image recognition to analyze pedestrian behavior.
• Trained and tested multiple ML models: decision trees, multilayer perceptrons, Bayesian algorithms, and Support Vector Machines (SVM).
• Extracted authentic pedestrian crossing data from signalized intersections in Chinese cities.

Main Results:

• The Support Vector Machine (SVM) model demonstrated superior accuracy.
• SVM effectively predicted pedestrian crossing probabilities and speeds.
• Identified SVM as optimal for pedestrian crossing prediction and traffic simulation.

Conclusions:

• Machine learning significantly enhances pedestrian crossing predictions.
• SVM is a highly effective model for intelligent transportation safety applications.
• This research contributes to safer, more efficient urban mobility through AI.