Real-Time Decoding of Movement Onset and Offset for Brain-Controlled Rehabilitation Exoskeleton
2026-03-17 • Robotics
RoboticsArtificial IntelligenceHuman-Computer Interaction
AI summaryⓘ
The authors developed a way for people to start and stop a robotic arm exoskeleton just by imagining movements, using brain signals measured non-invasively. Eight participants successfully controlled the robot to begin and halt reaching motions with reasonable accuracy during testing sessions. They also identified and fixed a common problem in the brain signal processing method that caused bias, improving the system's reliability and consistency. This work moves closer to practical robot therapy that responds directly to a patient's intentions, which might better support recovery after brain injuries.
robot-assisted therapymotor imageryEEGupper-limb exoskeletonneuroplasticitybrain-computer interfacesignal processingclass recenteringtask-specific trainingintent-driven control
Authors
Kanishka Mitra, Satyam Kumar, Frigyes Samuel Racz, Deland Liu, Ashish D. Deshpande, José del R. Millán
Abstract
Robot-assisted therapy can deliver high-dose, task-specific training after neurologic injury, but most systems act primarily at the limb level-engaging the impaired neural circuits only indirectly-which remains a key barrier to truly contingent, neuroplasticity-targeted rehabilitation. We address this gap by implementing online, dual-state motor imagery control of an upper-limb exoskeleton, enabling goal-directed reaches to be both initiated and terminated directly from non-invasive EEG. Eight participants used EEG to initiate assistance and then volitionally halt the robot mid-trajectory. Across two online sessions, group-mean hit rates were 61.5% for onset and 64.5% for offset, demonstrating reliable start-stop command delivery despite instrumental noise and passive arm motion. Methodologically, we reveal a systematic, class-driven bias induced by common task-based recentering using an asymmetric margin diagnostic, and we introduce a class-agnostic fixation-based recentering method that tracks drift without sampling command classes while preserving class geometry. This substantially improves threshold-free separability (AUC gains: onset +56%, p = 0.0117; offset +34%, p = 0.0251) and reduces bias within and across days. Together, these results help bridge offline decoding and practical, intention-driven start-stop control of a rehabilitation exoskeleton, enabling precisely timed, contingent assistance aligned with neuroplasticity goals while supporting future clinical translation.