Boston Dynamics Founder Shares Story of Bio-inspiration
By Amy DerBedrosian
Marc Raibert hasn’t yet created the robotic equivalent of the mountain goat or the pack mule, but he’s getting a lot closer. After more than 25 years, he believes development of a legged robot with the mobility and load capacity to be useful is near.
“I’ve more excited in the past couple of years than ever before,” says the founder of Boston Dynamics whose involvement with this effort began in the late 1970s.
Raibert’s tale of his lengthy quest to build machines with as much mobility and load carriage as the impressively rugged animals was the first presentation in a new monthly seminar series from UC Berkeley’s Center for Interdisciplinary Bio-Inspiration in Education and Research (CIBER). Like Raibert, CIBER is interested in merging biological principles and innovative engineering. CIBER director and professor of integrative biology Robert Full’s insights about the gecko that could lead to remarkable new adhesives exemplify how the center seeks discoveries about nature to inspire creative uses in industry.
Full, who also collaborated with Raibert on the robot project in the 1990s, said the guest speaker had over the years shown the “importance and beauty of dynamic-legged locomotion.”
Raibert’s first robot, though rudimentary, was a breakthrough. He recalled, “It was both an inspiration and a launching point for me. I was so excited by this robot that could move and walk.”
But the early robot moved statically and lacked dynamic locomotion. To improve it, Raibert decided to take a biological approach, examining the movements of animals. He said, “They are remarkable, dynamic systems, able to do their own sensing and control. That was my bio-inspiration.”
This led to a one-legged hopping robot with springs, followed by a two-legged version able to do a somersault. Raibert then moved on to a four-legged machine that could trot and gait and bound, built on control systems that focused on support, balance, and posture. Bouncing on springy legs, the robot could keep its body upright and move with symmetry.
After an eight-year hiatus from the project beginning in 1995, the self-described “roboholic” returned with a more ambitious goal: to develop a robot rugged enough for real-world applications.
The new-and-improved model is known as “Big Dog,” named for its design drawn from study of the joints of dogs. A 235-pound creation with a hip, knee, foot, ankle, actuator, leg spring, force sensor, engine-driving hydraulic system, and on-board computer, Big Dog can carry up to 340 pounds. It also has been put to the test on rough terrain and in extreme conditions, such as snow and ice. Big Dog’s performance continues to improve, and the robot has successfully completed up to a six-mile hike without refueling.
What remains for a practical Big Dog — what Raibert called a “plausibly useful robot” — is the ability to handle still rougher terrain, right itself, operate more reliably and quietly, and to function with greater autonomy.
“We don’t have the benefit of strong muscles and strong bones that biology can take advantage of. I wish we had something as strong as muscle, as energetic as muscle, but we just don’t,” Raibert said. “Instead, we just have a lot of power built into it.”
To learn more about CIBER and future events in its seminar series, visit: http://ciber.berkeley.edu/twiki/bin/view/CIBER/ProgramRegistration
.
