Body mobility is not something most of us think about too much. But as people age or need to learn to cope with a disability, mobility becomes a big concern. The simple, everyday activities that normally fill our lives become too difficult, and people with reduced mobility may find themselves socially isolated and unable to meet their own basic needs. Traditional mobility aids, such as walkers and wheelchairs, can be helpful, but these tools can also be very cumbersome and limit the types of activities one can participate in. What most people would like is a little boost when they need it. , without having to lug around big, heavy equipment.
Exoskeletons offer the promise of giving that extra boost when needed, but they’ve generally been very expensive and completely impractical for use outside of a lab. A recent paper published by a research group at Stanford University described a boot-like exoskeleton which can offer assistance to the wearer while walking, but it is also made from relatively inexpensive components and has been validated to work in real-world conditions. By using machine learning and off-the-shelf hardware components, this exoskeleton avoids the tethering and complicated setup typically required for such devices, and can provide real-world help to real people.
The device is completely autonomous (📷: K. Hickman)
The boot consists of a Raspberry Pi 4 single board computer, inexpensive sensors that monitor the walking movements of the wearer, a motor driver and a motor. A rechargeable battery is worn around the waist like a belt, making the exoskeleton completely self-contained – all power and processing units are worn on the body. The system applies torque to the ankle to partially replace the work normally done by the calf muscle. During a step, just before the toes leave the ground, the motor helps lift off the ground. This has a similar effect to removing a 30-pound backpack, according to the team.
As you can imagine, everyone has a somewhat different way of walking and the assistance each person needs is unique. Simply applying the same nudge to every step for every user would create a weird experience that is more likely to cause trouble than provide help. The exoskeleton manages this using a machine learning model that learns over time how best to assist the wearer. When a new user wears the boot for the first time, it takes about an hour of monitoring their walking habits for the device to adjust the pattern and customize it to provide the optimal level of assistance.
In actual testing of the boot, it was found that people could walk 9% faster while expending 17% less energy for a given distance walked compared to using normal shoes. This is a huge improvement over existing exoskeletons – it’s about double the reduction in walking effort seen with current state-of-the-art devices. Equally important, this new exoskeleton can actually be used outside of the lab and without a team of researchers to manually tune it. There is a lot of potential in this work to help people in real life situations.
The next step for the team is to clinically validate their device with older users and people with disabilities. They also plan to build variations of the device designed to reduce joint pain or help with balance. Once that work is complete, they hope to team up with commercial partners to bring the exoskeleton to market.