Scholarlink

<p>With the continuous advancement of global rehabilitation technologies, home-based lower-limb rehabilitation exoskeletons have demonstrated broad application prospects due to their low cost and high convenience. However, significant inter-patient variability, the complex and dynamic nature of the rehabilitation process, and the time-varying characteristics of knee joint torque present major challenges for these exoskeletons in accurately matching torque demands during training. Designing scientifically-grounded, personalized rehabilitation programs and precisely regulating training intensity in home settings remains a highly complex and challenging task. To address this challenge, we propose a home-use knee health monitoring system (KHMS) grounded in embodied intelligence, featuring three key innovations: (i) a four-bar variable-stiffness mechanism that provides adjustable stiffness over 10–40 N·m/rad; (ii) a self-powered electromagnetic module that delivers a stable 3.4 V output to reliably energize low-power wireless sensors; and (iii) an LSTM-based adaptive monitoring model that estimates knee joint torque with 95.36% accuracy and supports continuous rehabilitation-state tracking. By enabling real-time assessment and data-driven personalization based on individual recovery trajectories, the proposed KHMS facilitates scientifically grounded home training with precise intensity regulation. Overall, this work advances practical, portable rehabilitation devices that can meet community-level rehabilitation needs and promote the development of low-power artificial intelligence.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Structural Design and Performance Analysis of a Deep-Learning-Based Adaptive Variable-Stiffness Lower-Limb Rehabilitation Exoskeleton

  • 期刊论文

  • 发表日期：

    2026年2月24日

查看全文
  • Yingnan Wang,
  • Chenxi Wang,
  • Zhixia Wang,
  • Boxun Tao,
  • Wei Wang

摘要

With the continuous advancement of global rehabilitation technologies, home-based lower-limb rehabilitation exoskeletons have demonstrated broad application prospects due to their low cost and high convenience. However, significant inter-patient variability, the complex and dynamic nature of the rehabilitation process, and the time-varying characteristics of knee joint torque present major challenges for these exoskeletons in accurately matching torque demands during training. Designing scientifically-grounded, personalized rehabilitation programs and precisely regulating training intensity in home settings remains a highly complex and challenging task. To address this challenge, we propose a home-use knee health monitoring system (KHMS) grounded in embodied intelligence, featuring three key innovations: (i) a four-bar variable-stiffness mechanism that provides adjustable stiffness over 10–40 N·m/rad; (ii) a self-powered electromagnetic module that delivers a stable 3.4 V output to reliably energize low-power wireless sensors; and (iii) an LSTM-based adaptive monitoring model that estimates knee joint torque with 95.36% accuracy and supports continuous rehabilitation-state tracking. By enabling real-time assessment and data-driven personalization based on individual recovery trajectories, the proposed KHMS facilitates scientifically grounded home training with precise intensity regulation. Overall, this work advances practical, portable rehabilitation devices that can meet community-level rehabilitation needs and promote the development of low-power artificial intelligence.