Getting people moving – Walking exoskeletons could mobilize disabled patients

Prof. Jessy Grizzle has long said that his work in robotics could one day be used to help the disabled. Now he and his group, alongside French company Wandercraft, are working to make that claim a reality in the form of walking exoskeletons.

Prof. Grizzle's group working on robotic exoskeleton Enlarge
Prof. Jessy Grizzle, Ayonga Hereid, Margaret Eva Mungai wearing a V-3 exoskeleton, and Omar Harib

Prof. Jessy Grizzle has long said that his work in robotics could one day be used to help the disabled. Now he and his group, alongside French company Wandercraft, are working to make that claim a reality in the form of walking exoskeletons.

PhD student Omar Harib, postdoctoral researcher Ayonga Hereid, and PhD student Eva Mungai spent four days in July working with Wandercraft in Paris. The company’s goal is to create an exoskeleton that will allow patients that are paralyzed from the waist down to walk upright, with a natural gait and the freedom to use their hands.

Wandercraft's Exoskeleton version 2 Enlarge
Prof. Jessy Grizzle's team developed algorithms for walking motions for Wandercraft's Version 3 exoskeleton, shown above. They were excited by the challenge and opportunity of working on the just-finished Version 4 exoskeleton when they arrived in Paris.

Exoskeletons are still a relatively new area of research, and many competing products require users to use crutches for stability. This forces the patients to put weight on their arms, making the exoskeleton fatiguing to use and possibly resulting in more health problems. The Wandercraft Exoskelton assures stability with some of the same mechanisms that keep Grizzle’s robot MARLO  upright – Grizzle’s smart algorithms position its legs to ensure balance, and keep it moving at the desired speed and in the desired direction.

So far, Grizzle’s team has developed algorithms for the exoskeleton’s walking motion, programming a walking gait that works best for the exoskeleton’s design and fitting that software into its larger control package.

The team tested their algorithms on simulated models of the device for eight months before the trip. In Paris, they introduced major upgrades to their mathematical methods that could make Wandercraft’s exoskeleton safer for users and offer better performance. These upgrades came from Grizzle’s recently graduated PhD student Dennis Da, in his effort to get MARLO walking over the Wave Field.

It was Grizzle’s goal for the team to test their algorithms on the real device before they left Paris. With only four days to make it happen, the group worked under a lot of pressure.

“Given that we didn’t get the new exoskeleton model until day two, the goal seemed insane,” Grizzle says. “But I’ve learned that team Michigan tends to thrive with crazy goals. It added drama to the last afternoon.”

By 6:00pm on the last day of testing, the team began implementation. Half an hour later, the exoskeleton was walking in place, stable, with an upright posture.

This was the group’s first chance to apply their models to legged machines that attach to humans and assist them in their daily lives.

While an important step in the lab’s collaboration with Wandercraft, the trip was also a major opportunity for the team members themselves. This project will be the focus of Omar Harib’s PhD research, who was primarily responsible for adapting the team’s software package to the exoskeleton and planning its walking motion.

“Working on this project allows me to tackle the type of difficult problems I’m used to encountering with bipedal walking, while also having a more direct impact on human lives,” says Harib.

Hereid developed a key optimization algorithm as a PhD student at Georgia Tech. The package, called Fast Robot Optimization and Simulation Toolbox (FROST), helps the roboticists quickly prototype a system and develop algorithms to plan different types of motion. Hereid’s first accomplishment at Michigan was to turn FROST into a generalized framework that could be used with different robots.

“A key to the software was to make it easily adaptable to different circumstances,” says Hereid.

The adaptability of FROST allowed the team to adjust quickly when they were surprised with a new version of the exoskeleton upon their arrival – they were able to change their model and program a new walking gait in just two days.

For Eva Mungai, this trip was a trial-by-fire introduction to Grizzle’s team. Beginning her master’s in mechanical engineering this fall, she had just started her work a month before departure.

“Every incoming student has to start from scratch and code the simplest walking robot there is,” Mungai says. That robot is the three-link walker, a simple torso with two legs. “By doing that, I could understand the language that’s used in robotics. So when I went to France, I wasn’t totally clueless.”

This project is a step toward Mungai’s dream occupation of enhancing people’s lives with robotics – her personal statement was all about building an Ironman suit.

“They showed us videos of clinical trials, and you can see the smile on their face when they actually stand,” Mungai says. “It’s great.”

The team’s work on the exoskeleton so far has been published in their paper, “First Steps Toward Translating HZD Control of Bipedal Robots to Decentralized Control of Exoskeletons.” In their continuing work on the project, they will extend their focus to the exoskeleton’s control software.

NSF NRI Grant #1525006 supported this work, which is a joint effort with ECE alum  Prof. Koushil Sreenath, ME UC Berkeley, and Grizzle’s long time collaborator, Prof. Aaron Ames, ME, Cal Tech. The work of J. Grizzle was also partially supported by a gift from Ford Motor Company.