Harvard Engineers Develop Smart Shoulder Robot to Restore Arm Movement in Stroke and ALS Patients

Harvard researchers have developed a personalized control system for a soft wearable shoulder robot that significantly improves arm mobility in individuals with stroke or Amyotrophic lateral sclerosis (ALS). 

Physiotherapist assisting a patient with Amyotrophic Lateral Sclerosis.

Study: Personalized ML-based wearable robot control improves impaired arm function. Image Credit: David Herraez Calzada/Shutterstock.com

The system combines a machine learning model to detect movement intention with a physics-based model to enhance control accuracy and responsiveness. When tested on users with upper limb impairments, the system improved movement precision, expanded joint range of motion, and reduced compensatory motions—resulting in meaningful functional gains.

Why This Matters

Wearable robots are increasingly being explored as assistive tools for individuals with upper limb impairments. However, making these devices intuitive, reliable, and effective for daily use remains a challenge.

Current systems often rely on surface electromyography (sEMG), which is sensitive to electrode placement and can be inconsistent across users or over time. Additionally, many control systems don’t adequately model the relationship between robotic force and limb movement—especially in soft robots, where that relationship can be nonlinear and delayed.

Evaluations typically focus on isolated joint performance, leaving questions about whole-arm movement quality and real-world usability unanswered.

This study takes a different approach: it combines machine learning and physics-based modeling in a way that adapts to each user. The result is a soft robotic system that not only responds intelligently to intention but also moves with the body in a natural, supportive way.

How the Study Worked

The research involved nine participants with significant arm motor impairments—five of whom had experienced strokes and four diagnosed with ALS. The study followed a multi-day protocol using a soft wearable shoulder robot enhanced by two key components:

  1. An Intention Detection Model (IDM): A neural network trained to classify the user’s intent (to lift, hold, or lower the arm) in real time, using data from inertial measurement units (IMUs) and soft compression sensors.
  2. A Hysteresis Model: A physics-based system that calculated the precise amount of pneumatic pressure needed to support the arm’s movement based on its current angle and dynamic history.

On day one, the IDM was trained on each participant’s motion data. On day two, the full system was tested during a series of guided in-lab tasks—such as joint movements and reaching activities—with and without robotic assistance. High-resolution motion capture was used to quantify improvements in movement quality, including range of motion, trunk compensation, and hand-path efficiency.

To assess the robot’s practicality beyond the lab, two participants continued testing on a third day at home, performing everyday tasks like eating, drinking, and a timed weight-holding challenge. This real-world evaluation helped demonstrate the system’s adaptability and robustness in unstructured environments.

This wearable robot can learn

What They Found

The results were compelling. When the robot was active:

  • Shoulder, elbow, and wrist range of motion increased
  • Trunk compensation decreased
  • Hand movement became smoother and more efficient

The intention detection model achieved 94.2 % accuracy in real-time classification, while the hysteresis model enabled the system to reduce the force needed to move the arm by 31.9 %—helping users perform movements more comfortably and naturally.

At-home testing confirmed that these improvements held up during real-world activities, with participants completing functional tasks more easily than they could without the robot. The system didn’t just work in the lab—it proved helpful in everyday life.

Why it Matters Now

This study marks a step forward in wearable robotics by showing how a personalized, adaptive control system can improve upper-limb function in people with severe motor impairments. By fusing machine learning and biomechanical modeling, the system offers more intuitive support and more effective rehabilitation potential than previous approaches.

And because it’s designed with real-world use in mind—using robust sensors, practical training methods, and a lightweight soft robotic design—it could be viable for long-term use in both clinical and home settings.

Journal Reference

Arnold et al. (2025). Personalized ML-based wearable robot control improves impaired arm function. Nature Communications16(1). DOI:10.1038/s41467-025-62538-8. https://www.nature.com/articles/s41467-025-62538-8

Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.

Source:

Citations

Please use one of the following formats to cite this article in your essay, paper or report:

  • APA

    Nandi, Soham. (2025, August 27). Harvard Engineers Develop Smart Shoulder Robot to Restore Arm Movement in Stroke and ALS Patients. AZoRobotics. Retrieved on August 27, 2025 from https://www.azorobotics.com/News.aspx?newsID=16158.

  • MLA

    Nandi, Soham. "Harvard Engineers Develop Smart Shoulder Robot to Restore Arm Movement in Stroke and ALS Patients". AZoRobotics. 27 August 2025. <https://www.azorobotics.com/News.aspx?newsID=16158>.

  • Chicago

    Nandi, Soham. "Harvard Engineers Develop Smart Shoulder Robot to Restore Arm Movement in Stroke and ALS Patients". AZoRobotics. https://www.azorobotics.com/News.aspx?newsID=16158. (accessed August 27, 2025).

  • Harvard

    Nandi, Soham. 2025. Harvard Engineers Develop Smart Shoulder Robot to Restore Arm Movement in Stroke and ALS Patients. AZoRobotics, viewed 27 August 2025, https://www.azorobotics.com/News.aspx?newsID=16158.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this news story?

Leave your feedback
Your comment type
Submit

Sign in to keep reading

We're committed to providing free access to quality science. By registering and providing insight into your preferences you're joining a community of over 1m science interested individuals and help us to provide you with insightful content whilst keeping our service free.

or

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.

Read the full Terms & Conditions.