Brain-computer interfaces (BCIs) based on motor imagery (MI) are powerful tools for robotic platform control. However, the reliability of this control is highly influenced by the user's cognitive state. In particular, mental fatigue can affect the EEG signal, leading to a degradation in system performance. Many datasets exist for either MI or mental fatigue analysis, however they are typically studied independently, limiting the development of systems that can jointly interpret user intent and cognitive condition. This paper introduces a novel EEG dataset and a reproducible experimental protocol that integrates MI tasks with synchronized fatigue-level annotations. Using Open-BCI, PsychoPy, Lab Streaming Layer (LSL) protocol, and the LabRecorder, the acquisition system captures 16 EEG channels alongside a subjective (Karolinska Sleepiness Scale (KSS) ratings) and objective (reaction-time tasks) fatigue assessment. The protocol involves 5 participants over at least four sessions, each lasting 50 minutes, to capture inter-subject and inter-session variability. This integrated dataset structure is designed to support the development of multitask learning and cognitive-aware adaptive control in BCI-based robotic systems. The preliminary validation results of the proposed dataset are also reported.
A Novel EEG Dataset for Joint Motor Imagery and Mental Fatigue Analysis in BCI applications
Carissimo, C.;Cerro, G.;
2026-01-01
Abstract
Brain-computer interfaces (BCIs) based on motor imagery (MI) are powerful tools for robotic platform control. However, the reliability of this control is highly influenced by the user's cognitive state. In particular, mental fatigue can affect the EEG signal, leading to a degradation in system performance. Many datasets exist for either MI or mental fatigue analysis, however they are typically studied independently, limiting the development of systems that can jointly interpret user intent and cognitive condition. This paper introduces a novel EEG dataset and a reproducible experimental protocol that integrates MI tasks with synchronized fatigue-level annotations. Using Open-BCI, PsychoPy, Lab Streaming Layer (LSL) protocol, and the LabRecorder, the acquisition system captures 16 EEG channels alongside a subjective (Karolinska Sleepiness Scale (KSS) ratings) and objective (reaction-time tasks) fatigue assessment. The protocol involves 5 participants over at least four sessions, each lasting 50 minutes, to capture inter-subject and inter-session variability. This integrated dataset structure is designed to support the development of multitask learning and cognitive-aware adaptive control in BCI-based robotic systems. The preliminary validation results of the proposed dataset are also reported.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
