Research will lead to bio-inspired robots


Professor Phil Holmes: 'You don't want to slavishly copy biology, you want to be inspired by biology'

by Peter Page

Phil Holmes has achieved a great deal in his career.

A professor of mechanics and applied mathematics in the Department of Mechanical and Aerospace Engineering, Professor Holmes' renown as a mathematician has been recognized with his appointment, at different times, as director of the Center for Applied Mathematics at Cornell and the Program in Applied and Computational Mathematics here at Princeton, and appointment as an associate faculty member in the mathematics department.

He is a fellow of the American Academy of Arts and Sciences and has published more than 180 papers and four books, as well as four collections of poetry.

Yet somewhere on his crowded résumé it ought to be noted that Professor Holmes has proved that every creature on Earth has its charms if you get to know it well enough--even the cockroach.

"When you get to studying these things, they get quite attractive," he said, after an enthusiastic recounting of the cockroach's ability to run with amazing speed and unerring aim toward safety, despite being all-but brainless.

He is making a significant contribution to the knowledge needed to build ever-more useful robots, though a self-described "incompetent" robot builder.

Among his other research interests, Professor Holmes collaborates with Robert Full, founder of the Poly-PEDAL Laboratory at the University of California at Berkeley. Dr. Full's laboratory specializes in studying how nature's superb engineering of insects, crabs, and lizards allows those creatures, despite their tiny brains, to ingeniously navigate over, under, and around obstacles.

"To climb stairs or get over rocks or fallen beams, such as at the World Trade Center after the collapse, you need legs," Professor Holmes said. "You'd like to have nimble, agile robots that operate in a dynamic environment. The current legged robots operate in a static environment and move slowly, while the wheeled robots can move fast but are limited to smooth surfaces. Insects have been getting around extremely well on legs, and rather fast, too, for a lot longer than we have."

Cockroaches, with their zippy speed and uncanny ability to dodge aside just as a boot heel is about to squash them are lab favorites, though crabs and geckos have been studied with good results. Research at the lab has led to (pardon the pun) great strides in robot locomotion.

"My motivation primarily is understanding nature, not building robots," Professor Holmes said. "I'm interested in anything that moves, evolves, or changes in space or time."

Professor Holmes' collaborators at the Poly-PEDAL Lab have done exhaustive work to discover the secrets of how cockroaches race around with such amazing skill. Professor Holmes has translated that tsunami of data into a mathematical model.

"I'm the theorist here," he said. "I don't do the experiments, I don't build the robots. I am trying to put together the fundamental understanding that can be used in many other instances."

In terms of locomotion, the most advanced robots are to a cockroach what a cart is to a Porsche.

Cockroaches have elaborate sensory feedback systems, such as force transducers on their legs and hair cells that measure velocities. It will be a while before anybody can build a robot as swift and nimble as a cockroach, though RHex, a six-legged machine built by Martin Buehler at the Center for Intelligent Machines at McGill University, using some of the principles Professor Holmes has discovered, is running in the right direction.

In the meantime, Professor Holmes believes it more useful to understand the cockroach in its simplicity instead of all its complexity.

"To me, a lot of science is ignoring effects that are small and trying to focus on the key underlying principles," he said. "Unlike many of my colleagues in engineering, I don't want to include everything. I want to exclude as much as I can. To understand basic principles it is best to go to a very simple model."

A cockroach runs on its six legs with a gait in which the front, back, and opposite middle legs are on the ground, giving it a tripod of support.

To get a more applicable understanding of that, Professor Holmes, working with graduate students John Schmidt, Raffaele Ghigliazza, and Justin Seipel, started with bipedal models having a single leg on the ground at any time.

"These simple models show us that, depending upon where we pivot our single effective leg, it will or will not be stable," Professor Holmes said. "We can describe them by as few as four parameters. This is ridiculously simple, but it really helps understanding."

Professor Holmes' work serves as a bridge between biologists and engineers building robots. The bridge is mathematical theory that extracts the key underlying principles of the cockroach's mobility.

"That helps people build a machine that is fast and nimble, without bothering with everything that is irrelevant," he said.

"You don't want to slavishly copy biology, you want to be inspired by biology. These simple mathematical models nail down biological inspiration so it can be immediately translated into the mechanical world."