Commercial knee orthoses are widely used in rehabilitation, providing passive support, lateral stability, and safety limits to prevent post-surgical tendon re-rupture. However, conventional orthoses lack access to kinematic and kinetic information, motivating sensor integration for objective assessment and motorization for assistance-as-needed (AAN) therapy. This paper presents the sensorization and motorization of a commercially available knee orthosis with a modular design, preserving its clinical usability. An open-source ROS2 framework enables real-time hardware–software integration, scalable control architecture, and simultaneous data acquisition. A transparency control strategy, based on a velocity disturbance observer, compensates induced disturbances, ensuring smooth, low-impedance motion. The actuator, motor encoder, and inertial measurement unit (IMU) provide real-time data for gait analysis, detecting key phases such as mid-swing, heel contact, and toe-off. Experimental evaluation demonstrates that the disturbance observer outperforms the LuGre feedforward friction model in tracking friction dynamics under increasing amplitude and frequency conditions, as confirmed by lower RMSE and MAE metrics. Integration within ROS2 allows high-frequency operation (800 Hz) while executing parallel tasks. This work demonstrates that an existing clinical orthosis can be effectively upgraded to support data-driven rehabilitation strategies, providing a foundation for AAN and resistive training. Future developments include gravity compensation and machine-learning-based activity assessment to further enhance patient-centered therapy.
Just et al. (Wed,) studied this question.