United States Department of Transportation
i
Investigation of Driver Adaptations in a Mixed Traffic Environment
-
2024-09-01
[PDF-1.72 MB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Report
-
Corporate Publisher:
-
Abstract:This study investigates the impact of semi-automated vehicle (SAV) systems, specifically Adaptive Cruise Control (ACC), on driver behavior and control transitions (CT). The variations and adaptations in driver performance due to mental workload, influenced by factors such as task difficulty and driver awareness, are critical, especially under different driving conditions. As ACC systems often fail to provide the necessary deceleration in sudden critical traffic situations, a transition from automated to manual control is triggered. This research aims to predict these CTs using ensemble machine learning (ML) models and to analyze the key factors contributing to these transitions using SHAP analysis. Driving data from 30 participants in both manual and ACC conditions were collected using a driving simulator, including variables such as vehicle trajectories, driver demographics, and mental workload. Various scenarios, including vehicle cut-ins, merging, and lane drops, were developed to capture driver reactions and build predictive models. Among all the ML models, XGBoost produced the best overall performance with accuracy, F1 score, and ROC_AUC values of 0.75, 0.83, and 0.76 respectively. Additionally, SHAP analysis was performed to explore the prominent factors behind the CTs. The study finds significant differences in driving behavior between manual and ACC conditions, with ensemble ML models providing robust predictions of CTs. The findings suggest that age, experience, relative velocity, and perceived mental workload are the key factors behind CT. The results underscore the importance of enhancing ACC systems to improve driver safety and comfort, particularly in critical traffic scenarios.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
