Wearable robots for everyday life: Harvard aims to make walking aids smarter after stroke

At the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS), Professor Conor Walsh's group develops soft wearable robots that physically assist movement, as well as sensing systems that give feedback on how people move. In a conversation published on 7 October and an accompanying video from the school, Walsh describes how data analysis and machine learning are meant to tune the devices more closely to the people wearing them.

In a project with Patrick Slade (SEAS) and Lou Awad and Terry Ellis of Boston University, the team uses sensors already built into the wearable robot to estimate walking speed, leg symmetry and push-off. The system then adjusts its control to encourage a more natural gait after a stroke. Another project with Susan Murphy explores when a smartwatch should give stroke survivors feedback at home on how much they use their arm.

The aim is to obtain movement data from a few sensors in everyday life that is nearly as accurate as lab measurements – for example on how Parkinson's disease affects walking. To that end, the group is building datasets that pair lab measurements with wearable-sensor data. When predictions are uncertain, a device should switch to a more conservative mode. "Doing no harm is more important than trying to do as much good as possible," Walsh says. The piece does not report study results with figures.