Engineers have designed robots that crawl, swim, fly and even slither like a snake, however no robotic can maintain a candle to a squirrel, which may parkour via a thicket of branches, leap throughout perilous gaps and execute pinpoint landings on the flimsiest of branches.
College of California, Berkeley, biologists and engineers are attempting to treatment that state of affairs. Primarily based on research of the biomechanics of squirrel leaps and landings, they’ve designed a hopping robotic that may stick a touchdown on a slim perch.
The feat, to be reported within the March 19 challenge of the journal Science Robotics, is an enormous step within the design of extra agile robots, ones that may leap among the many trusses and girders of buildings beneath development or robots that may monitor the setting in tangled forests or tree canopies.
“The robots we now have now are OK, however how do you are taking it to the subsequent stage? How do you get robots to navigate a difficult setting in a catastrophe the place you could have pipes and beams and wires? Squirrels may try this, no downside. Robots cannot try this,” mentioned Robert Full, one in every of paper’s senior authors and a professor of integrative biology at UC Berkeley.
“Squirrels are nature’s finest athletes,” Full added. “The best way that they’ll maneuver and escape is unbelievable. The concept is to attempt to outline the management methods that give the animals a variety of behavioral choices to carry out extraordinary feats and use that data to construct extra agile robots.”
Justin Yim, a former UC Berkeley graduate scholar and co-first creator of the paper, translated what Full and his biology college students found in squirrels to Salto, a one-legged robotic developed at UC Berkeley in 2016 that might already hop and parkour and stick a touchdown, however solely on flat floor. The problem was to stay the touchdown whereas hitting a selected level — a slim rod.
“If you concentrate on attempting to leap to a degree — perhaps you are doing one thing like taking part in hopscotch and also you need to land your ft in a sure spot — you need to stick that touchdown and never take a step,” defined Yim, now an assistant professor of mechanical science and engineering on the College of Illinois, Urbana Champaign (UIUC). “If you happen to really feel like you are going to fall over ahead, you then may pinwheel your arms, however you will additionally most likely arise straight so as to preserve your self from falling over. If it feels such as you’re falling backward and also you might need to sit down down since you’re not going to have the ability to fairly make it, you may pinwheel your arms backward, however you are probably additionally to crouch down as you do that. That’s the similar conduct that we programmed into the robotic. If it will be swinging beneath, it ought to crouch. If it will swing over, it ought to prolong out and stand tall.”
Utilizing these methods, Yim is embarking on a NASA-funded mission to design a small, one-legged robotic that might discover Enceladus, a moon of Saturn, the place the gravity is one-eightieth that of Earth, and a single hop may carry the robotic the size of a soccer area.
The brand new robotic design is predicated on a biomechanical evaluation of squirrel landings detailed in a paper accepted for publication within the Journal of Experimental Biology and posted on-line Feb. 27. Full is senior creator and former graduate scholar Sebastian Lee is first creator of that paper.
Mixing biology and robotics
Salto, brief for Saltatorial Agile Locomotion on Terrain Obstacles, originated a decade in the past within the lab of Ronald Fearing, now a Professor within the Graduate Faculty in UC Berkeley’s Division of Electrical Engineering and Pc Sciences (EECS). A lot of its hopping, parkouring and touchdown potential is a results of a long-standing interdisciplinary collaboration between biology college students in Full’s Polypedal Lab and engineering college students in Fearing’s Biomimetic Millisystems Lab.
Throughout the 5 years Yim was a UC Berkeley graduate scholar — he bought his Ph.D. in EECS in 2020, with Fearing as his adviser — he met with Full’s group each different week to be taught from their biology experiments. Yim was attempting to leverage Salto’s potential to land upright on a flat spot, even outside, to get it to hit a selected goal, like a department. Salto already had a motorized flywheel, or response wheel, to assist it steadiness, a lot the way in which people wheel their arms to revive steadiness. However that wasn’t enough for it to stay a direct touchdown on a precarious perch. He determined to attempt reversing the motors that launch Salto and use them to brake when touchdown.
Suspecting that squirrels did the identical with their legs when touchdown, the biology and robotics groups labored in parallel to substantiate this and present that it might assist Salto stick a touchdown. Full’s group instrumented a department with sensors that measured the power perpendicular to the department when a squirrel landed and the torque or turning power with respect to the department that the squirrel utilized with its ft.
The analysis group discovered, based mostly on high-speed video and sensor measurements, that when squirrels land after a heroic leap, they mainly do a handstand on the department, directing the power of touchdown via their shoulder joint in order to emphasize the joint as little as doable. Utilizing pads on their ft, they then grasp the department and twist to beat no matter extra torque threatens to ship them over or beneath the department.
“Nearly the entire power — 86% of the kinetic power — was absorbed by the entrance legs,” he mentioned. “They’re actually doing entrance handstands onto the department, after which the remainder of it follows. Then their ft generate a pull-up torque, if they are going beneath; if they’re going to go excessive — they’re overshooting, probably — they generate a braking torque.”
Maybe extra necessary to balancing, nevertheless, they discovered that squirrels additionally alter the braking power utilized to the department when touchdown to compensate for over- or undershooting.
“If you are going to undershoot, what you are able to do is generate much less leg-breaking power; your leg will collapse some, after which your inertia goes to be much less, and that can swing you again as much as right,” Full mentioned. “Whereas if you’re overshooting, you need to do the other — you need to have your legs generate extra breaking power so that you’ve got a much bigger inertia and it slows you down so that you could have a balanced touchdown.”
Yim and UC Berkeley undergraduate Eric Wang redesigned Salto to include adjustable leg forces, supplementing the torque of the response wheel. With these modifications, Salto was capable of leap onto a department and steadiness a handful of occasions, even though it had no potential to grip with its ft, Yim mentioned.
“We determined to take essentially the most tough path and provides the robotic no potential to use any torque on the department with its ft. We particularly designed a passive gripper that even had very low friction to attenuate that torque,” Yim mentioned. “In future work, I believe it might be fascinating to discover different extra succesful grippers that might drastically develop the robotic’s potential to regulate the torque it applies to the department and develop its potential to land. Possibly not simply on branches, however on advanced flat floor, too.”
In parallel, Full is now investigating the significance of the torque utilized by the squirrel’s foot upon touchdown. Not like monkeys, squirrels would not have a usable thumb that enables a prehensile grasp, so they need to palm a department, he mentioned. However which may be a bonus.
“If you happen to’re a squirrel being chased by a predator, like a hawk or one other squirrel, you need to have a sufficiently secure grasp, the place you’ll be able to parkour off a department rapidly, however not too agency a grasp,” he mentioned. “They do not have to fret about letting go, they simply bounce off.”
One-legged robots might sound impractical, given the potential for falling over when standing nonetheless. However Yim says that for leaping actually excessive, one leg is the way in which to go.
“One leg is the most effective quantity for leaping; you’ll be able to put essentially the most energy into that one leg if you happen to do not distribute that energy amongst a number of completely different units. And the drawbacks you get from having just one leg reduce as you leap larger,” Yim mentioned. “Whenever you leap many, many occasions the peak of your legs, there’s just one gait, and that’s the gait by which each leg touches the bottom on the similar time and each leg leaves the bottom at roughly the identical time. So at that time, having a number of legs is form of like having one leg. You may as effectively simply use the one.”
Different co-authors of the Science Robotics paper are Fearing and former UC Berkeley undergraduate Eric Wang, now a graduate scholar at MIT, and former graduate scholar Nathaniel Hunt, now an affiliate professor on the College of Nebraska in Omaha. Co-authors of the J. Exp. Bio. paper are Wang, Hunt, Fearing, UC Berkeley Affiliate Professor of Mechanical Engineering Hannah Stuart and former UC Berkeley undergraduates Stanley Wang and Duyi Kuang. The analysis was funded by the U.S. Military Analysis Workplace (W911NF-18-1-0038, W911NF-1810327) and the Nationwide Institutes of Well being (P20GM109090).