Lane-keeping ability evaluation for driving skill tests: A multi-indicator fusion approach

Area of Science:

• * Road Safety and Driver Behavior Analysis
• * Human-Computer Interaction in Transportation

Background:

• * Traditional driver skill tests often fail to capture real-world driving performance, leading to suboptimal outcomes.
• * Lane-keeping ability is crucial for driver competence but is difficult to assess effectively in standardized tests.
• * Existing methods lack comprehensive evaluation of lane-keeping due to complex influencing factors.

Purpose of the Study:

• * To develop and validate a novel multi-indicator fusion (MIF) method for assessing driver lane-keeping ability in skill tests.
• * To enhance the accuracy and real-world relevance of driver skill evaluations.
• * To provide a framework for assessing lane-keeping in future autonomous driving contexts.

Main Methods:

• * Extraction of multidimensional lane-keeping indicators from low-speed naturalistic driving data, focusing on lateral and longitudinal safety and stability.
• * Application of K-means clustering to group indicators with similar characteristics.
• * Determination of indicator thresholds using Youden index, Boxplot, and statistical measures, followed by Analytic Hierarchy Process (AHP) for a comprehensive evaluation model.

Main Results:

• * The proposed MIF method effectively extracts and integrates multiple indicators to evaluate lane-keeping ability.
• * Threshold determination using statistical measures enhanced evaluation accuracy.
• * The AHP-based model provided a rational and feasible approach validated by naturalistic driving data.

Conclusions:

• * The MIF method offers a more accurate and realistic assessment of driver lane-keeping ability compared to traditional tests.
• * This approach provides valuable insights for improving driver skill testing and evaluating autonomous driving systems.
• * The study establishes a robust framework for assessing a critical aspect of driving competence.