Hybrid Systems to Boost EEG-Based Real-Time Action Decoding in Car Driving Scenarios

Vecchiato, Giovanni · 2021 · Crossref

DOI: 10.3389/fnrgo.2021.784827

archive: archived pipeline: cataloged verified

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

Summary

This perspective paper addresses the challenge of decoding real-time driving actions, such as braking and steering, using electroencephalography (EEG). Driving is a complex neuroergonomic task involving simultaneous cognitive, motor, and attentional processes. While EEG offers high temporal resolution for tracking these cerebral dynamics, its application in real-world scenarios is limited by low signal-to-noise ratios, artifacts from muscle and eye movements, and high inter-subject variability. These factors hinder the reliability of unimodal EEG-based brain-computer interfaces (BCIs) for assistive driving technologies. The paper proposes hybrid systems that combine EEG with peripheral physiological signals, such as electromyography (EMG) and electrooculography (EOG), to enhance prediction accuracy, reduce false positives, and improve the ecological validity of action decoding. The author reviews existing literature on hybrid BCI systems in driving contexts, analyzing studies that utilized both simulated and real-world driving environments. The review highlights methodological approaches where EEG captures early neural correlates of motor preparation, while EMG and EOG provide robust markers of movement execution. Specific studies examined include those using contingent negative variation (CNV) potentials to predict braking and steering up to 320–800 ms before action onset. The paper also details recent work by Vecchiato et al. (2021), which combined non-ecological EEG data (elicited by traffic sign cues) with ecological EMG data (collected during driving simulation) to predict steering direction. This approach leveraged the early predictive power of EEG to boost the classification performance of EMG signals, which are easier to acquire in natural settings. Key findings indicate that hybrid systems significantly outperform unimodal approaches in both accuracy and timing. EEG allows for earlier prediction of intent compared to EMG, which is more accurate but occurs closer to movement onset. Combining these signals enables the detection of emergency braking approximately 300 ms before onset in real-world scenarios and steering actions up to 1.5 s in advance when using cued tasks. The integration of EEG features with EMG activity improved the discrimination of steering sides during simulation, demonstrating that neural data collected offline can enhance the predictive power of peripheral signals collected online. The review confirms that hybrid architectures mitigate the limitations of raw EEG data, such as non-stationarity and artifact susceptibility, by incorporating complementary physiological information. The significance of this work lies in advancing the technological maturity of neuroergonomic applications. By validating hybrid systems in driving scenarios, the paper establishes a framework for developing assistive devices that can predict driver behavior with sufficient lead time to intervene safely. The authors suggest that this methodology can extend beyond driving to other high-risk domains, such as occupational safety and telerehabilitation, where early action prediction is critical. The ability to use less invasive peripheral sensors in ecological settings, informed by neural data, offers a practical pathway for deploying brain-based monitoring systems in daily life, potentially reducing accidents and improving user support in complex operational environments.

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

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