Need a Lift?

Everybody needs a little help sometimes. And those with impaired mobility, perhaps due to a neurodegenerative disease or stroke, might need a bit more specialized assistance. A wearable robot developed at Harvard University holds a great deal of promise for these individuals, especially those with upper-limb impairments. It gives assistance to those who still have control over their arms, but not enough strength to raise them long enough to do their normal daily chores, like brushing their teeth or eating meals.

The wearable takes the form of a vest. It works by tracking the motion of the wearer then inflating balloons underneath the arms with air to aid in lifting them when needed. While this device can give a big boost to those that need it, the Harvard-based researchers soon found that the design was suboptimal. Everyone has different body mechanics and needs different types of assistance, so a one-size-fits-all approach like this can’t quite cut it in all cases.

Based on their observations, the researchers recently made some updates to the vest. They added pressure sensors so that the device would not only know that assistance was needed, but how much assistance to give. Moreover, they incorporated a machine learning model and a physics-based algorithm into the control system to tailor every movement to the user’s unique needs.

These updates turned the vest from something that feels like it is pushing and pulling at the body into something that feels more like a natural extension of the body. The device relies on a system of inertial measurement units, which detect changes in motion, along with the newly added compression sensors. These sensors capture the subtle physical cues of when the user is trying to lift or lower an arm. The machine learning model then interprets the data to identify the user’s motion intention with a high level of accuracy.

One of the early problems with the system was that once the robotic vest helped lift an arm, some users struggled to bring it back down. To address this, the researchers built in a physics-based hysteresis model that estimates the minimum amount of pressure needed to support an arm, and then dynamically adjusts as the user moves. The combination of learning-based personalization and physics-guided transparency allows the robot to provide exactly the right level of support, helping when it is needed, and getting out of the way when it is not.

Testing was conducted with nine volunteers: four living with ALS and five recovering from strokes. Compared to earlier prototypes, participants using the upgraded vest experienced greater ranges of motion in their shoulders, elbows, and wrists. They also required about a third less force to lower their arms, making daily activities like eating, drinking, or brushing hair more comfortable. In addition, users showed fewer compensatory movements, such as leaning or twisting their torsos to achieve a task, meaning their motions were more efficient and natural.

The technology remains in development, but the team sees a lot of potential. Stroke patients might benefit from the vest as part of rehabilitation, while people with progressive conditions like ALS may rely on it more for daily assistance. At present, the researchers are continuing to work toward making the system portable and usable at home.