Classification of Driver Steering Intentions Using an Electroencephalogram

IKENISHI, Toshihito; KAMADA, Takayoshi; NAGAI, Masao · 2008 · Crossref

DOI: 10.1299/jsdd.2.1274

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 addresses the development of a brain-computer interface (BCI) for driver assistance systems, aiming to classify a driver’s steering intentions using electroencephalogram (EEG) signals. The motivation stems from the need for cooperative driving assistance systems that adapt to individual driver preferences and intentions, rather than relying solely on vehicle dynamics. By decoding brain activity, the system seeks to determine operational intent before physical action occurs, enabling more responsive and personalized safety interventions. The researchers conducted experiments with five able-bodied male subjects of varying driving experience using a driving simulator. Subjects performed a task involving avoiding other vehicles by steering left, right, or continuing straight. EEG data was recorded from nine scalp electrodes (F3, F4, C3, C4, P3, P4, Fz, Cz, Pz) at 1 kHz sampling rate. The experimental paradigm involved a 2-second cognition/judgment period after a visual cue indicated the required steering direction, followed by a 2-second execution period. The classification algorithm consisted of preprocessing, feature extraction, and classification. Preprocessing involved band-pass filtering (0.1–100 Hz), notch filtering (50 Hz), and a sharpness filter to enhance signal-to-noise ratio. Features were extracted using Fast Fourier Transformation (FFT) over 1-second windows, focusing on frequency bands from 6–31 Hz. The classification module utilized a multivariate normal distribution to model feature densities for four states: "stand by," "straight running," "right turn," and "left turn." Bayesian decision theory was applied to calculate the posterior probability of each intention, incorporating state transition probabilities via a finite state automaton. A probability threshold of 0.75 was used to confirm classifications, with uncertain results defaulting to "stand by." The off-line analysis revealed an average classification accuracy of approximately 65% across all subjects. Performance varied by driving experience; subjects with frequent driving experience (Subjects A and B) achieved accuracies of 70–80%, while occasional drivers showed lower consistency. Subject E, a novice, demonstrated high accuracy for left turns but poor performance for straight driving. The study noted that classification errors often occurred when the feature distribution deviated from the assumed normal distribution or when the 1-second window included pre-cue data. The results indicate that while EEG-based intention classification is feasible, the current probabilistic model and feature extraction methods require refinement to improve reliability and reduce misclassification rates. This work contributes to the field of human-machine interfaces by demonstrating the potential of non-invasive brain signal processing for real-time driving assistance applications.

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.

• steering pattern
• drowsiness detection algorithms

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
• Methodological Resource: tool software
• Theoretical Contribution: computational model