Stroke is one of the leading causes of non-traumatic disability worldwide, affecting more than 15 million people each year, with about three-quarters experiencing long-term functional impairments. This makes it crucial to develop long-term rehabilitation programs that can promote motor relearning, enhance neural plasticity, and restore daily motor function. Robot-assisted rehabilitation, which combines neuroscience, biomechanics, and advanced control systems, is emerging as a highly promising approach.
In recent years, exoskeleton-type rehabilitation robots that enable distributed interaction across multiple joints have gained popularity due to additional improvements in upper-limb motor abilities. Some prominent examples of such robots include ArmeoPower, ANYexo, and EXO-UL8. They employ control strategies that optimize robotic intervention and maximize active patient participation. However, few existing systems simultaneously incorporate motion intention detection, desired trajectory generation, and assistance level personalization for effective and efficient neurorehabilitation.
In an innovation, a team of researchers from China, led by Professor Zeng-Guang Hou from the State Key Laboratory of Multimodal Artificial Intelligence Systems (MAIS), Institute of Automation, Chinese Academy of Sciences, have developed CASIA-EXO, an upper-limb exoskeleton, and proposed an exciting control strategy for motor learning in post-stroke patient-in-the-loop neurorehabilitation. Their novel findings were published in Volume 12, Issue 8 of the IEEE/CAA Journal of Automatica Sinica on 20 August 2025. The team includes researchers Chen Wang and Liang Peng from MAIS.
According to Prof. Hou, “CASIA-EXO is a five degree-of-freedom biomimetic exoskeleton that comprises three rotational joints adopting an oblique arrangement and two rotational joints co-locating in a serial chain. Its dynamics is modelled and linearized to identify unknown parameters embedded in the shoulder, elbow and wrist mechanisms.”
Following the modelling, the researchers propose a novel patient-in-the-loop control strategy for rehabilitation training for post-stroke motor recovery.
“It consists of the intention-based trajectory planning and performance-based intervention adaptation. While an oscillator-based intention estimator quantifies the time-varying training requirement and integrates the invariant laws in normal motion patterns into the multi-joint trajectory generation, the performance-based adaptive algorithm alters the assistance and resistance levels during the robot-aided rehabilitation, which can enhance active participation of subjects with various impairment levels,” explains Dr. Wang.
The research team conducted various experiments to demonstrate the efficacy of their proposed system, wherein 10 healthy subjects sat in a chair with their dominant arm coupled with CASIA-EXO and passed the wooden boxes over and then brought them back using a virtual reality display. They found that their novel control strategy steadily individualized the training trajectory and intervention level as per the subject’s changing requirements and motor abilities, facilitating closed-loop robot-aided rehabilitation. Consequently, the close cooperation between trajectory planning and intervention adaptation can aid motor relearning and functional recovery in patients with motor impairments as naturally as possible.
This breakthrough underscores the potential of CASIA-EXO to deliver safer, smarter, and more personalized rehabilitation for stroke survivors.
***
Reference
DOI: 10.1109/JAS.2024.124662
About the Chinese Academy of Sciences
The Chinese Academy of Sciences (CAS) is the national academy for science and technology in China. It is the nation’s comprehensive center for research and development in the natural sciences, comprising more than 100 research institutes. CAS drives scientific innovation and provides guidance to the Chinese government and the world. Comprising a comprehensive research and development network, a merit-based learned society, and a system of higher education, it brings together scientists and engineers from China and around the world to address both theoretical and applied problems using world-class scientific and management approaches.
Website: http://english.cas.cn/
About Professor Zeng-Guang Hou from the Institute of Automation, Chinese Academy of Sciences
Zeng-Guang Hou is a Professor at the State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, China. He is also associated with the University of Chinese Academy of Sciences and the CAS Center for Excellence in Brain Science and Intelligence Technology. His research interests include computational intelligence, medical robots, and intelligent systems.
About Associate Professor Chen Wang from the Institute of Automation, Chinese Academy of Sciences
Chen Wang is an Associate Professor at the State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, Beijing. She received her Ph.D. in control theory and control engineering from the University of Chinese Academy of Sciences in 2021. Her research focuses on human–machine interaction, and intelligent rehabilitation robotics.
About Associate Professor Liang Peng from the Institute of Automation, Chinese Academy of Sciences
Liang Peng is an Associate Professor at the State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, China. His research interests include rehabilitation, haptics, and robotics.
About IEEE/CAA Journal of Automatica Sinica
The IEEE/CAA Journal of Automatica Sinica publishes high-impact research across all areas of automation, serving as a global forum for innovative ideas and advancements. With an Impact Factor of 19.2 (ranked first worldwide, top 0.6%) and a CiteScore of 28.2 (TOP 2 in “Control and Optimization”), it leads the field. As of 2025, its real-time SCI Impact Factor is 11.93, with 62 ESI top papers (TOP 1%) and 6 hot papers (TOP 0.1%). Strengthened collaboration with IEEE and global platforms continues to expand its international influence.
Funding information
This work was supported in part by the National Key Research and Development Program of China (2022YFC3601200), the National Natural Science Foundation of China (62203441, U21A20479), the Beijing Natural Science Foundation (L232005), and the Inner Mongolia Autonomous Region Science and Technology Plan (2023YFDZ0042).
END