A Multimodal Sensor-Integrated Lower Limb Exoskeleton With Digital Twin Framework for Real-Time Gait Rehabilitation

This study proposes a lower-limb exoskeleton rehabilitation system that integrates a proprietary multimodal sensor network with a real-time digital twin framework to enhance mobility training and gait recovery. The system incorporates a hybrid sensor array composed of piezoelectric sensors fabricated via near-field electrospinning (NFES) and piezoresistive sensors, enabling accurate acquisition of gait biomechanics and center of pressure (COP) dynamics. A field-oriented control (FOC) system in conjunction with a controller area network (CAN) protocol facilitates real-time data exchange between sensing and actuation modules. A sensor fusion algorithm combines plantar force-distribution data with real-time virtual reality (VR) feedback, thereby supporting adaptive gait training. The digital twin module generates a synchronized virtual replica of the user’s lower-limb motion, enabling detailed gait analysis and motion tracking. Furthermore, an embedded Fuzzy-COP control algorithm dynamically adjusts assistive torque based on individual biomechanical characteristics, enabling personalized rehabilitation. This integrated platform advances precision rehabilitation by converging real-time sensing, intelligent control, and immersive feedback technologies. The proposed Fuzzy-COP method improves gait-phase classification accuracy by 6.7% over threshold-based methods and maintains VR-exoskeleton synchronization with root-mean-square error (RMSE) < 1.3°. The NFES sensors also exhibit < 3% drift after one million cycles, confirming long-term stability.