Comparative Analysis of Robotic Assistive Devices on Paretic Knee Motion in Post-Stroke Patients: An IMU-Based Pilot Study

<b>Background</b>: Robotic assistive devices are increasingly used in post-stroke gait rehabilitation, yet quantitative evaluations of synchronization between human and robotic joint motion remain limited. This study examined gait kinematics in post-stroke hemiplegic patients using two exoskeleton-type devices-HAL^® (Cyberdine Inc., Tsukuba, Japan) and curara^® (AssistMotion Inc., Ueda, Japan)-based on synchronized IMU measurements. <b>Methods</b>: Two post-stroke patients performed treadmill walking under non-assisted and assisted conditions with HAL^® and curara^®. Only the paretic knee joint was analyzed to focus on the primary control joint during gait. Inertial measurement units (IMUs) simultaneously recorded human and robotic joint angles. Synchronization was assessed using Bland-Altman (BA) analysis, root mean square error (RMSE), and mean synchronization jerk (MSJ). The study was designed as an exploratory methodological case study to verify the feasibility of synchronized IMU-based human-robot joint measurement. <b>Results</b>: Both assistive devices improved paretic knee motion during gait. RMSE decreased from 7.8° to 4.6° in patient A and from 8.1° to 5.0° in patient B. MSJ was lower during curara-assisted gait than HAL-assisted gait, indicating smoother temporal coordination. BA plots revealed reduced bias and narrower limits of agreement in assisted conditions, particularly for curara^®. Differences between HAL^® and curara^® reflected their distinct control strategies-voluntary EMG-based assist vs. cooperative gait-synchronization-rather than superiority of one device. <b>Conclusions</b>: Both devices enhanced synchronization and smoothness of paretic knee motion. curara^® demonstrated particularly smooth torque control and consistent alignment with human movement. IMU-based analysis proved effective for quantifying human-robot synchronization and offers a practical framework for optimizing robotic gait rehabilitation. The novelty of this study lies in the direct IMU-based comparison of human and robotic knee joint motion under two contrasting assistive control strategies.