Traffic Sign Recognition Evaluation for Senior Adults Using EEG Signals

Area of Science:

• Gerontology
• Cognitive Neuroscience
• Transportation Safety

Background:

• Elderly drivers face challenges recognizing road signs due to age-related cognitive decline and vision changes like presbyopia.
• Common road sign shapes (circles, squares, triangles) were used to assess recognition accuracy in older adults.
• Head-up displays (HUD) were employed to present traffic symbols for comparison with external signs.

Purpose of the Study:

• To evaluate the road sign recognition abilities of elderly drivers compared to younger individuals.
• To analyze event-related potential (ERP) data from EEG signals to differentiate correct and incorrect responses.
• To explore the effectiveness of machine learning techniques in classifying cognitive responses to traffic signs.

Main Methods:

• A Go/Nogo test was administered, measuring EEG signals and event-related potentials (ERP).
• Subjects recognized traffic symbols presented on a HUD.
• Unsupervised and supervised deep learning models were applied to EEG data for classification.

Main Results:

• Elderly drivers exhibited a 1.5 times higher error rate in road sign recognition compared to younger drivers.
• All age groups showed a 20-30 ms P300 delay in ERP for incorrect answers.
• Supervised deep learning achieved 75% accuracy in classifying elderly drivers' data, outperforming unsupervised methods.

Conclusions:

• Elderly drivers have significantly lower road sign recognition accuracy.
• Deep learning models, particularly supervised approaches, show promise for analyzing cognitive responses in older drivers.
• Findings support the development of personalized safe driving systems for the elderly.