Related Experiment Video
Updated: Nov 10, 2025
11:12
Driving Simulation in the Clinic: Testing Visual Exploratory Behavior in Daily Life Activities in Patients with Visual Field Defects
Published on: September 18, 2012
17.6K
Investigating the correspondence between driver head position and glance location
Joonbum Lee1, Mauricio Muñoz1,2,3, Lex Fridman1
1AgeLab and New England University Transportation Center, Massachusetts Institute of Technology, Cambridge, MA, United States of America.
Peerj. Computer Science
|April 5, 2021
Summary
Driver head pose accurately estimates gaze direction, especially for larger shifts. This technology can help detect driver distraction and inattention, improving road safety.
Area of Science:
- Human-Computer Interaction
- Automotive Safety
- Machine Learning
Background:
- Driver gaze estimation is complex due to individual and contextual variations.
- Understanding the link between head pose and glance orientation is crucial for driver monitoring.
Purpose of the Study:
- To investigate head pose as a reliable estimator for driver gaze direction.
- To analyze the accuracy of predicting glance locations using head rotation data.
Main Methods:
- Statistical analysis correlating head rotation with manually coded gaze data.
- Machine learning models, including Hidden Markov Models (HMM), for gaze classification.
- Evaluating classification accuracy based on visual angles between glance locations.
Main Results:
- Classification accuracy improved with increased visual angles between glance locations.
- Hidden Markov Models achieved 83% accuracy in distinguishing forward roadway glances from center stack glances.
- Head rotation data shows potential robustness to individual differences in head-glance correspondence.
Conclusions:
- Driver head pose can serve as a surrogate for eye gaze, particularly for high-eccentricity glances.
- Head pose tracking offers a cost-effective method for developing driver distraction and inattention detection systems.
