Hybrid BCI Replaces Exoskeleton Crutch Controls at 95.06% SSVEP Accuracy
Summary
A hybrid brain-computer interface replaced crutch control for 10 participants walking with a custom lower-limb exoskeleton, classifying steady-state visual evoked potentials with 95.06% accuracy and reaching F1 scores of 99.80% and 99.22% for its wink- and clench-triggered asynchronous switches, researchers reported in IEEE TNSRE. Measured against crutch control, the system scored 78.25 versus 53.50 on the System Usability Scale and 3.15 versus 7.85 on NASA-TLX physical demand.
Why it matters
Exoskeleton control research often reports decoding accuracy from seated subjects, which says little about a system a user must operate while walking. Reporting SSVEP accuracy and switch F1-scores under gross body movement, alongside a head-to-head usability comparison against crutch control, is the more informative framing. Sample size is small at 10 participants, and the cohort is not described as a patient population.
BCIwiki (bciwiki.com) – The study was published in IEEE Transactions on Neural Systems and Rehabilitation Engineering on September 1, 2026. The team developed a hybrid brain-computer interface combining a steady-state visual evoked potential (SSVEP)-based BCI with asynchronous biosignal-based switches triggered by a wink and a teeth clench, offered as an alternative to the crutch-based control conventionally used for lower-limb exoskeletons. Crutch-based control often imposes a considerable physical burden and limits usability, the authors write.Practical usability was improved with a wearable headband-type biosignal-recording device that acquires electroencephalography, electromyography and electrooculography signals, while augmented reality glasses present visual stimuli and gait guidance information, according to the paper. In the proposed asynchronous operational framework, SSVEP responses handle movement-mode selection, and the wink- and clench-based switches are assigned to command execution and cancellation. Ten participants completed real-time walking experiments wearing a custom lower-limb exoskeleton under both the conventional crutch-based method and the proposed one, with performance measured by classification accuracy and switch F1-scores and with usability and workload assessed using the System Usability Scale (SUS) and the NASA Task Load Index.Despite gross body movement during exoskeleton-assisted walking, the hybrid control framework reached an average SSVEP classification accuracy of 95.06%, with F1-scores of 99.80% and 99.22% for the wink- and clench-based switches respectively, the authors say. Against crutch-based control, the method scored significantly higher on the System Usability Scale, 78.25 versus 53.50, and showed significantly lower physical demand on the NASA Task Load Index, 3.15 versus 7.85. The evidence does not state whether the participants were a patient population.
Compiled by BCIwiki from public sources
