A Real-Time Recognition System of Driving Propensity Based on AutoNavi Navigation Data

Area of Science:

• Traffic Safety
• Intelligent Transportation Systems
• Machine Learning Applications

Background:

• Driving propensity, reflecting driver attitude and behavior, is crucial for traffic safety and accident reduction.
• Current methods for assessing driving behavior lack real-time, personalized insights.
• AutoNavi navigation data offers a rich source for analyzing dynamic driving patterns.

Purpose of the Study:

• To propose a novel real-time system for identifying driving propensity using AutoNavi navigation data.
• To develop and validate a machine learning model for accurate driving propensity assessment.
• To facilitate the creation of personalized intelligent driver assistance systems.

Main Methods:

• A dynamic data acquisition method was developed to collect time, speed, and acceleration from AutoNavi navigation.
• Principal Component Analysis (PCA) was employed for feature extraction from experimental vehicle data.
• A Fruit Fly Optimization Algorithm (FOA) combined with Generalized Regression Neural Network (GRNN) was utilized to build the FOA-GRNN model.

Main Results:

• The dynamic data acquisition and analysis methods were validated through real vehicle experiments.
• Principal Component Analysis effectively extracted key driving propensity characteristics.
• The FOA-GRNN model demonstrated a high overall accuracy of 94.17% in identifying driving propensity.

Conclusions:

• A practical and accurate real-time driving propensity identification system was successfully constructed and validated.
• The proposed method offers a convenient approach for developing personalized intelligent driver assistance systems.
• This research contributes to improving traffic safety through advanced data analysis and machine learning techniques.