Mental Fatigue Evaluation for Passive and Active BCI Methods for Wheelchair-Robot During Human-in-the-Loop Control

Human-Robot Interaction (HRI) and Brain-Computer Interfaces (BCIs) aim to develop practical and realistic solutions to assist both healthy and disabled individuals in their Activities of Daily Living (ADL). This paper focuses on the application of BCIs and Brain Machine Interfaces (BMIs) to correct or perform actions by machines through brain stimulation. The authors investigate mental load and fatigue during extended tasks using mental load indices in BCI human-in-the-loop experiments, specifically involving human observation and error perception to rectify errors in obstacle detection by a wheelchair-mobile robot. The study compares the passive method utilizing Error-Related Potentials (ErrPs) with potential active methods such as P300 or Steady-State Visually Evoked Potential (SSVEP) in terms of mental load and signal classification accuracy. The Mental Fatigue Index (MFI) and other power-ratio indices are used to estimate mental load. Preliminary results indicate that passive BCIs induce lower mental fatigue compared to active methods for the same task, making them suitable for applications involving elderly or disabled individuals. However, the passive method utilizing ErrPs exhibits lower task engagement, causing participants to lose attention and feel drowsy, leading to reduced classification accuracy compared to other BCI methods such as SSVEP.