Physip à l’ICCAS 2024 – Vendredi 17 mai à 15h – Monitoring mental workload with a single EEG channel

Physip sera présente à l’International Conference on Cognitive Aircraft Systems (ICCAS) qui se déroulera les 16 et 17 mai 2024 à Toulouse sur le campus de l’ISAE-SupAero. Christian Berthomier, PDG de Physip, donnera un talk vendredi 17 mai à 15h, en clôture de la session parallèle Sensors for crew monitoring :

Monitoring mental workload with a single EEG channel, Christian Berthomier, Gaëtan Sanchez, Romain Bouet, Emmanuel Maby, Julien Mattei, Youri Dmitriev, Marie Brandewinder, Anne-Isabelle Mallet Dacosta, Régis La Combe, Jérémie Mattout

Objective and significance
Staff, whether civilian or military, are integrated into increasingly complex systems. Anticipating when they will be overwhelmed is crucial to the smooth running of operations. Cognitive overload presents operational risks since it coincides with a decline in performance, encompassing aspects such as diminished understanding, misinterpretation of circumstances, ineffective communication and tunneling. Therefore, preserving operational efficiency largely depends on avoiding cognitive overload scenarios. However, to achieve this, such situations must first be recognized.
The DGA/AID project COGNISIM aimed at developing an algorithm to monitor mental workload and optimize simulation-based military training, in the context of tank turret shooting.

Methods
The algorithm is based on the analysis of brain activity captured by the electroencephalography (EEG). In order to facilitate the transition of EEG from controlled laboratory environments to practical, real-world applications, the automatic analysis is single-channel based (Cz-Pz).
We acquired 117 recordings from 61 volunteers who performed two 1h30 protocols, on a computer and on a turret simulator. In addition to EEG (16 channels), signals electrocardiogram, electrodermal activity and pupillometry were recorded. The subject self- assessment of workload, as well as the one provided by the instructor, were logged. The subject accuracy and performance were also logged.
Pupillometry and EEG data analysis were first compared in their ability to distinguish between the ON/OFF-task states. They were also compared according to their relation with behavioral performance and task difficulty.

Discussion
Our preliminary results validate the feasibility of monitoring mental workload with a minimal EEG setup. Preliminary results indicate that while EEG demonstrates comparable performance to pupillometry in a computer task, it surpasses it in a tank turret simulator scenario. This superiority likely arises because EEG is rather insensitive to variations in brightness, contrary to pupillometry.

Research funded by the General Directorate for Armament, Ministry of the Armies. DGA/AID/RAPID Cognisim project 172906081.

> Télécharger l’abstract