EEG-Based Detection of Braking Intention Under Different Car Driving Conditions

Hernández, Luis G.; Mozos, Oscar Martinez; Ferrández, José M.; Antelis, Javier M. · 2018 · Crossref

DOI: 10.3389/fninf.2018.00029

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates the feasibility of detecting emergency braking intention from driver electroencephalographic (EEG) signals prior to mechanical brake pedal actuation. The research is motivated by the high mortality rate of traffic accidents, largely attributed to human error and delayed reaction times exacerbated by cognitive states such as stress, workload, and fatigue. While existing driver assistance systems rely on external sensors, they often fail to account for the driver’s internal cognitive state. By identifying brain activity associated with braking intention, advanced systems could anticipate emergency maneuvers earlier, potentially reducing accident rates. The experimental design involved seven right-handed male participants driving in a simulator using the Open Racing Car Simulator software. Participants followed a guide vehicle that performed unexpected emergency braking, requiring the driver to react to avoid collision. To simulate real-world conditions, the study induced stress via ambulance sirens, workload via a simultaneous attention task, and fatigue by conducting sessions in the afternoon following daily activities. EEG signals were recorded from seven electrodes (F3, F4, C3, C4, Cz, P3, P4) and electromyographic (EMG) signals from the right leg. Data preprocessing included filtering and segmentation into pre-response epochs (preceding brake pedal deflection) and normal driving epochs. The researchers employed two classification algorithms to discriminate between braking intention and normal driving: Support Vector Machines (SVM) and Convolutional Neural Networks (CNN). Time-domain features were extracted from the EEG signals, dividing each 1.5-second epoch into ten 150-ms intervals. The CNN architecture consisted of two convolution-pooling layers followed by a feed-forward artificial neural network. The study analyzed 480 emergency braking situations per participant across eight experimental combinations of cognitive factors. The results demonstrated significant recognition of emergency braking intention. The average classification accuracy was 71.1% for SVM and 71.8% for CNN. Individual performance varied, with the best participant achieving 80.1% accuracy using SVM and 88.1% using CNN. These findings confirm that EEG signals contain distinguishable patterns preceding the physical act of braking, even under varying cognitive loads. The study concludes that incorporating driver bioelectrical signals into advanced driver-assistance systems is feasible for the early detection of emergency braking situations, offering a pathway to mitigate accidents caused by delayed human reactions.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• drowsiness detection algorithms
• braking response

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: physiological data, behavioral performance data
• Theoretical Contribution: computational model