January 26, 2015
Topic: Campus News
Despite tremendous advances in the field of two-legged robots during the past few decades, bipedal machines are a long way from impersonating, much less improving upon, the human gait.
In his inaugural lecture as the Elmer G. Gilbert Distinguished University Professor of Engineering, Jessy Grizzle will discuss the efforts underway in his lab to close this gap.
Distinguished University Professor is the highest professorial title granted at U-M. Grizzle's lecture, "Taking Bipedal Walking Robots from Science Fiction to Science Fact" is at 4 p.m. Feb. 4 in Rackham Amphitheatre. A reception will follow in the Assembly Hall.
Grizzle is the Jerry W. and Carol L. Levin Professor of Engineering, with appointments in electrical and computer engineering as well as mechanical engineering.
"In science fiction, robots walk, run, and jump better than you," Grizzle says. "In reality, can you count on them to walk over rubble and pull you from a burning building? Not so much."
For more than a decade, Grizzle has been working to make bipedal 'bots move like people. He and his team devised a nonlinear control theory of two-legged locomotion that enables their machines to do this better than their counterparts. Because the math behind the theory is based on the "nonlinear" nature of walking, their equations can reconcile both its continuous aspects — one foot in front of the other in front of the other — and the discrete parts — when a foot hits the ground and sends impact shocks into the system, Grizzle said. Other approaches can't do this.
Grizzle's group has been honing this theory across three generations of robots. Rabbit, retired in 2005, inspired a new family of gait-control algorithms that allowed it to walk at a specific speed and keep its balance through surprise shoves and across changes in terrain.
MABEL, the world's fastest two-legged robot with knees, could jog at a 9-minute-mile pace, step down from an unexpected 8-inch stair and navigate wooden planks in its path. It earned Grizzle and a colleague a spot on Popular Mechanics' 2012 list of 10 World Changing Innovators. MABEL was retired in 2013.
Both robots were connected to a boom in the lab and could only move in circles. MARLO, the newest platform, is designed to travel in free space. At this point, it can walk, but is easily tripped up, and Grizzle and his graduate students are working to make it stable enough to walking all over North Campus.
Other than their capability for grace, what sets these robots apart is how efficiently they carry themselves, Grizzle says. Boston Dynamics' Atlas, for example, is a 330-pound Titan that will compete in this year's finals of the Defense Advanced Research Projects Agency's Robotics Challenge. The aim of this Olympics for the metal set is to advance rescue robots.
Atlas, which Grizzle describes as "a very advanced" machine, uses between 20 and 100 times more energy than a human does. For MABEL and MARLO that factor is between two and three. Greater energy efficiency means longer battery life, and that could be important for the applications these robots are geared toward.
Two-legged robots could one day help in disaster situations like the Fukushima nuclear crisis. (That's what the DARPA challenge is modeled after.) They could also assist in the home or the factory.
"Ideally, we want machines that can function in human environments," Grizzle says. "So either we redesign the environment to allow quadrupeds and hexapeds to move around in it when we need help or we design machines that can work in the environment we already have."