Electrical Engineering and Computer Science

Meet the New Director of Michigan Robotics

  Bookmark and Share

Now making breakthroughs on his third robot, Jessy Grizzle is still somewhat amazed at where he ended up. He was originally a theory guy, content to let other engineers deal with the moving, and often breaking, parts.

Grizzle grew up on a farm in Oklahoma, far from the world of engineering. The first of his family to go to college, he found his way to Oklahoma State University, looking for a way to get his math and physics fix and hopefully land a job one day.

He fell in love with control theory as a PhD student at UT Austin, and says he would have been happy to remain a somewhat anonymous graduate student the rest of his life, tackling one intriguing problem after another.

Though he couldn’t remain a student, he did continue to solve problems – and he did so in spectacular ways that forced him into the limelight.

Jessy Grizzle, the Elmer G. Gilbert Distinguished University Professor and Jerry W. and Carol L. Levin Professor of Engineering, has been written about in Wired, his work featured on CNN and The Discovery Channel, and his robot videos viewed around the world – even in Michigan Stadium during the football games.

But though he appears in dozens of videos challenging robots to walk down impossibly steep slopes and through snow, some of his most-cited papers come from work on hybrid electric vehicles (HEVs). In fact, his relationship with the automobile, and Ford in particular, goes back 20 years, initiated because he wanted to be a better teacher.

I have often used my own failures to define my next challenges. My own experience helps me understand the freedom and community support that new faculty need to follow bold agendas that no one else believes in – and part of my job is to give it to them.

While introducing control to undergraduate students in his first job at the University of Illinois Urbana-Champaign, Grizzle quickly discovered that the students wanted more than just theory. “With zero street cred in practical applications, I knew I had to do something to change the situation,” he said.

So Grizzle threw himself into industrial work, beginning with a summer internship at Ford. That summer turned into two decades of summers. While there, to reach certain challenging company goals, he began to use model-based control and dynamic programming in order to improve fuel economy and reduce emissions. Moving away from hand turning of controller parameters and toward theoretically-based control software solutions seemed radical at the time, and until he proved it could work, there were plenty of doubters.

Toward the end of that period, during which he was named on 16 patents in the area of environmentally friendly powertrains, he turned to a technology that was still young – hybrid electric vehicles. Working with Huei Peng, a professor of mechanical engineering, Grizzle developed a power management control strategy for hybrid electric trucks. The algorithms described in the resulting paper are now part of the HEV technology used in FedEx trucks.

Grizzle’s current grant in cyber-physical systems has him tackling advanced driver assist programs such as adaptive cruise control and lane keeping, taking him into the space where the lines between car and robot start to blur.

“Cars are becoming big robots,” says Grizzle. “They may be the robot that we will have the most contact with in our daily lives.”

What are robots? “Whatever the faculty say they are,” says Grizzle. “The world of robotics is now so broad, it almost defies definition.”

His path to bipedal robotics began with a fortuitous sabbatical in France, where he encountered RABBIT.

“You’ve got to follow your passions. Don’t follow the trends. Don’t follow the money.”

At the time, RABBIT was just a concept for a bipedal, walking robot that was mechanically simpler and less costly than other robots of its kind. The team had all the know-how to design and build the machine, but no one knew how to give it a sense of balance and make it move with agility.

So Grizzle spent the following four years developing increasingly complex models of walking machines, starting from a stick figure, adding knees, and eventually generalizing the math to work with the novel RABBIT design.

When RABBIT took its first steps in July 2002, no other biped machine walked faster, was as stable, or varied its walking speed so adroitly. In fact, RABBIT walked on its first try — an unprecedented feat. The robot caught the eyes of the world, putting on a show in major publications and news services.

More than a decade later, Grizzle is now two robots (and two media blitzes) deeper into the field, showing off bipedal robots that are walking more nimbly than ever.

When MABEL began to run, she quickly earned the title of world’s fastest two-legged robot with knees. With no vision system, she saved herself from tripping over random terrain thanks only to the control principles and algorithms developed by Grizzle and his students.

“To me, MABEL was superhuman walking,” says Grizzle. “I think a person in her shoes would have face-planted.”

Her successor, MARLO, is Grizzle’s first biped to walk in 3D. Unbound from a lateral support boom, MARLO has wandered outside the lab, up and down hills, over rough terrain, and in snow. She’s even attempted the undulating Wave Field, and made a few improbable, yet successful, steps over the small mounds.

The algorithms behind Grizzle’s robots are public, making them applicable beyond the walls of his lab. They are now found in innovations across the nation, including groundbreaking new prosthetic limbs at the Chicago Rehabilitation Institute.

“My own students have the same freedom I enjoyed as a young person, and that’s why we’re going in directions I never dreamed of.”

Despite the attention these robots have brought to his lab, Grizzle is most comfortable working behind the scenes where he can blithely follow the thrill of a new challenge.

“You’ve got to follow your passions,” insists Grizzle. “Don’t follow the trends. Don’t follow the money.”

But you don’t have to go it alone if you don’t want to. The emphasis on collaboration in the planning of the new Robotics Building will ensure its success, according to Grizzle, who has collaborated with researchers at Michigan and around the world in many different fields.

As the inaugural Director of Robotics, he wants to enable the young faculty in particular to reach the high levels of success he sees in them.

“The people we’re hiring don’t need my advice,” says Grizzle. “They just need high-quality research space where they can dream, and the confidence to go ahead and take risks and even fail in the pursuit of those dreams. I have often used my own failures to define my next challenges. My own experience helps me understand the freedom and community support that they need to follow bold agendas that no one else believes in – and part of my job is to give it to them.”

And that goes for the students as well. “My own students have the same freedom I enjoyed as a young person,” Grizzle says, “and that’s why we’re going in directions I never dreamed of.”

But he’ll leave the tracks for others to follow. “The reason to do the mathematics is you get solutions that are good for not just one robot, but whole families of robots. And you leave roadmaps that others can follow to do things you never imagined.”


Posted: September 15, 2016