This study leverages a 14-channel Emotiv EEG system and NASA TLX to assess the cognitive load of wearable robot users during flooring tasks. By integrating brainwave data with subjective workload assessments, advanced machine learning algorithms were developed to accurately predict users’ cognitive load, paving the way for smarter and more adaptive robotic assistance in physically demanding environments.
Abstract
Active back-support exoskeletons are increasingly being perceived as potential solutions to the ergonomic risks of construction work. However, users of exoskeletons are susceptible to increased cognitive load could thwart the benefits of the device. Using self-reported cognitive load and electroencephalogram (EEG) data, this study investigated the detection of the cognitive load of users of an active back-support exoskeleton. EEG data and perceived ratings of cognitive load from participants performing flooring tasks are trained with several classifiers. The performance of the best classifier, Ensemble, improved using synthetic minority oversampling technique. This study contributes to existing knowledge by providing evidence of the extent to which cognitive load can be detected from the brain activity of exoskeleton users. The study also advances knowledge of the extent to which synthetic data could enhance the detection of cognitive load. Therefore, the study opens doors for improving exoskeleton designs to better support human cognition and performance.
Okunola, A., Akanmu, A., Jebelli, H. (2024). Detection of Cognitive Loads during Exoskeleton-Use for Construction Work. 2024 ASCE Construction Institute and Construction Research Congress Joint Conference, Des Moines, Iowa, US. https://doi.org/10.1061/9780784485293.088.
Akanmu, A., A., Afolabi, A., Okunola, A. (2023). Predicting mental workload of using exoskeletons for construction work: A deep learning approach. International Conference on Construction Applications of Virtual Reality. https://doi.org/10.36253/979-12-215-0289-3.69.
Methodology
