Autonomous Pediatric Exoskeleton Takes Its First Steps
Robotic Medical Device Advances Physical Therapy for Children
Once upon a time in northern France, Luc Masson, president of the INJENO Association, dreamed that his nine-year-old daughter would be able to walk. Inès has multiple neurological conditions that affect her motor skills, including epilepsy. She uses a wheelchair with a headrest and requires assistance to move through the world around her.
In 2010, Masson approached Laurent Peyrodie’s robotics and mechatronics group at the JUNIA HEI Graduate School of Science and Engineering, part of Lille Catholic University. He envisioned an autonomous lower limb exoskeleton that pediatric clinicians could use as a physical therapy tool. The device would also give children like his daughter a chance to experience the world in a new way.
“Luc is very enthusiastic and sympathetic,” remembered Peyrodie, who also leads Lille’s Biomedical Signal Processing Unit. “He asked, ‘Could you do something for us?’ With this question in mind, we started several different studies with students at the university.” Peyrodie recruited postdoctoral researcher Yang Zhang to head up mechanical and electronic development.
A colleague walking with an exoskeleton prototype. (Video credit: JUNIA HEI)
The JUNIA HEI Motion Project unites technicians, mechanical and biomedical engineers, and healthcare professionals to make mechanized orthosis a reality for children with severe neurological disorders. More than a dozen commercial, nonprofit, and academic partners joined the effort and the project secured €7.4 million ($7.2 million) in research funding from a European program.
Peyrodie and Zhang created prototypes from 3D-printed parts, low-vibration direct current motors, and embedded sensors. They accelerated development using MATLAB®, Simulink®, and Simscape Multibody™ to model motor dynamics and design motor controllers. The team turned to MathWorks partner Speedgoat for efficient real-time testing on a prototype hardware.
Recently, a petite colleague participating in the project walked in the latest prototype. In the coming months, the team plans to seek approval for testing the exoskeleton with children.
Taking the First Step
The Motion Project cited Swedish research that found that 30% of children in the country with cerebral palsy (CP) were unable to walk by age five. Another 16% utilized assistive devices to walk. Based on these statistics, the project team estimated that exoskeleton technology could benefit 6,500 children under age 10 in Europe. And that’s just for those with CP. There are many more with other disorders that affect leg movement.
“We needed to find a way to build the prototype in a short amount of time with just a few people.”
Despite this healthcare need, exoskeletons still tend to be viewed as gear for adults, such as the one that the fictional Iron Man dons or the real-world ones designed to prevent injury. Stateside, the U.S. Army developed the Human Universal Load Carrier (HULC) for lifting as much as 91 kilograms (200 pounds) and has plans to send a new flexible, lightweight exoskeleton into the field next year.
Making an exoskeleton that is approved for medical use for children with multiple disabilities, including many who experience difficulties in communicating, requires thorough clinical evaluation for safety. “Testing the structure on children in a clinical environment is extremely difficult because we have to prove that there is no risk or danger to them,” Peyrodie said.
Obtaining ethical approval from a committee is a crucial step, he continued. After that, the European Medicines Agency has a lengthy assessment process that medical device makers must go through before they can put a Conformité Européenne (CE) mark on a product.
Another challenge in pediatric lower limb exoskeleton development has been price. The thinking was that fast-growing children require different systems, which multiplies the cost. This is why the Motion Project’s unique system is designed to accommodate changing heights and weights.
The JUNIA HEI Motion Project is for eight- to 12-year-olds who have a broader range of disabilities, including lower limb paraplegia. The structure must be adjustable to wearers with varying heights who weigh up to 50 kilograms (110 pounds).
“This is quite a complicated project,” Peyrodie said. “It’s similar to a startup: We needed to find a way to build the prototype in a short amount of time with just a few people.”
Interreg 2 Seas, a European Territorial Cooperation program covering the area linked by the English Channel and the North Sea, invested in numerous sustainable and inclusive crossborder cooperation projects from 2014 to 2020. The program approved funding for the Motion Project in 2019. Although the original project timeline was extended slightly during the pandemic, the team still had to move quickly on prototyping.
Walking as One
Bionically reimprinting walking movement requires constructing a sophisticated robotic system robust enough for young humans and their caretakers to completely trust its autonomous functionality. The medical device also must be straightforward enough for a therapist to use safely.
Initially, the researchers tried designing their exoskeleton in C++, but the manual coding process dragged along. “That was impossible to manage in the time we had for the project,” Peyrodie recalled. “A project partner at KU Leuven in Belgium was using MATLAB and Simulink, combined with Speedgoat®, as a timesaving solution.”
The duo turned to Model-Based Design with MATLAB and Simulink. “MATLAB has mass calculations where you just tap the name of the function and then you can calculate it directly,” Zhang said. “Designing in C++, you have a lot of code to enter to realize one functionality. Model-Based Design with MATLAB and Simulink was really a timesaving tool for us.”
“Model-Based Design with MATLAB and Simulink was a timesaving tool for us.”
Model of the JUNIA exoskeleton. (Video credit: JUNIA HEI)
![The lower half of a mannequin wearing the prototype on one leg.](https://www.mathworks.com/company/mathworks-stories/autonomous-pediatric-exoskeleton-advances-physical-therapy-for-children/_jcr_content/mainParsys2/columns_663673900/835f3930-8ec3-4753-ad0a-8797c37eebc7/columns_320727814_co_2028947426/edfdebfc-8eb6-4efd-8edd-16e2771e95fa/image.adapt.full.medium.jpg/1699626446324.jpg)
An early exoskeleton prototype built with 3D-printed parts. (Image credit: JUNIA HEI)
Using Simscape Multibody, they constructed a dynamic model of their exoskeleton’s motors with a child-sized mannequin and applied future scenarios in the simulation environment. This allowed them to understand the parameters their exoskeleton motors require, narrowing the list of candidates. Peyrodie and Zhang ultimately chose Maxon brushless motors, in large part because the company’s drivers communicate via the controller area network (CAN) industrial protocol and integrate smoothly with the real-time open network EtherCAT.
Machining exoskeleton metal parts with computer numerical control from the start wouldn’t have allowed the researchers to make changes on the fly. Instead, the team ordered custom 3D-printed parts through an online marketplace starting with plastic for small early prototypes. Once they figured out which motors to use, the team felt confident graduating to full-sized metal components.
The researchers installed pressure sensors and a CAN bus in each foot plate along with inertial measurement units (IMUs). An additional IMU in the exoskeleton back enabled the team to measure posture. Altogether, the exoskeleton has six actuated degrees of freedom with mechanically limited ranges.
An autonomous exoskeleton won’t work without an effective communication system linking the motors and sensors. “We found that Simulink Real-Time™ is the best choice because you can design whatever model you want, whatever control algorithm you want, and you can apply it quickly to your prototype,” Zhang said. “It also uses a high-speed industrial communication protocol.”
Headquartered in Switzerland, Speedgoat produces MATLAB- and Simulink-compatible test systems, including a real-time target machine small enough for Peyrodie and Zhang to embed in the prototype and connect to their computer with an internet cable. Speedgoat’s system, which also has Maxon motor driver compatibility, meant the team could rapidly simulate and test their control designs.
Early in the pandemic, access to the university’s campus was limited. In-person work ceased, but the researchers were able to complete some simulation and design remotely. “We still developed,” Zhang said. “You always have small adjustments and modifications to make the prototype better.”
Exploring New Ground
The exoskeleton contains a sensory system that includes the inertial measurement units, encoders on the joints for controlling the angle, and a ground reaction force sensor. Four sensors in each exoskeleton foot detect where the human’s foot presses—important for maintaining balance.
MATLAB and Simulink cut their total prototyping time in half.
“The control algorithm we developed for weight transfer means that before starting to step, we transfer the mass to the front foot and then easily lift the free foot to move forward automatically,” Zhang said. “Because we have sensors on each foot, we know if the weight was transferred. If it’s successful, that will automatically trigger the next step.” Future plans include developing a control algorithm so the exoskeleton can lengthen steps when necessary, increasing stability.
Automatic stepping wasn’t in the original specification, but clinicians collaborating with the project team pointed out that humans don’t typically come to a complete stop between steps. Peyrodie said that the sensors and MATLAB made integrating that automation much easier than attempting to code it themselves.
“If we used C++ or C-Sharp development, we would need specialists in coding systems,” Peyrodie said. Zhang agreed, estimating MATLAB and Simulink cut their total prototyping time in half.
Peyrodie and Zhang also produced a user interface with MATLAB App Designer, enabling clinicians and partners to test the exoskeleton. Healthcare professionals’ feedback is vital, especially given the varying national approaches to rehabilitation. Peyrodie pointed out that physical therapists in France favor periodic limb stimulation, while in Poland they emphasize daily exercise. Many simply don’t have enough data for guidance.
More than Movement
New insights could be on the way. Belgian Motion Project partner Centexbel is working on a smart t-shirt with embedded sensors that measure body temperature, heart rate, and range of motion. Other partners are developing signal processing to detect stress level and a graphical user interface to display the data. The garment promises to be far less obtrusive than classic electrodes.
“We hope it will be used for rehabilitation training so that children can continue walking or regain that ability.”
“We want to be able to evaluate the child’s stress level while walking with our exoskeleton because it’s a major factor in whether or not this kind of system will be adopted,” Peyrodie said.
Their exoskeleton has passed electrical tests and is set for evaluation at a rehabilitation center in the Netherlands where clinicians have experience in robotic therapy for adults. Obtaining ethical approval takes a while, but Peyrodie said this stage is important for demonstrating the system’s efficiency and safety. If all goes well, a clinical trial with children will start this winter.
The team never forgot Luc Masson’s passionate plea.
Down the road, the project team sees the exoskeleton functioning in two ways. “We hope it will be used for rehabilitation training so that children can continue walking or regain that ability,” Zhang said. The other application is more emotional. “Like with Inès, she could psychologically sense the feeling of walking. And that might make her happy.”
Read Other Stories
AI / MEDICAL DEVICES
Combining Artificial Intelligence and Footwear Improves Diabetes Treatment
Smart Shoe for Diabetic Neuropathy Gets a Tech Upgrade
AI
Reconnecting the Brain After Paralysis Using Machine Learning
Brain-Computer Interface Restores Sense of Touch with Haptic Signals
AI
Cyborg Drummer and AI Team Create Music That’s Not Humanly Possible
Designing a Robotic Prosthetic for a Professional Drummer