Robotics

MIT's Mini Cheetah Robot can do a perfect backflip

MIT's Mini Cheetah Robot can do a perfect backflip
The MIT Mini Cheetah robot has a few tricks up its sleeve
The MIT Mini Cheetah robot has a few tricks up its sleeve
View 3 Images
The MIT Mini Cheetah robot can pick itself up when knocked down
1/3
The MIT Mini Cheetah robot can pick itself up when knocked down
The MIT Mini Cheetah robot has a few tricks up its sleeve
2/3
The MIT Mini Cheetah robot has a few tricks up its sleeve
The Mini Cheetah is far smaller than the latest version of the Cheetah robot
3/3
The Mini Cheetah is far smaller than the latest version of the Cheetah robot
View gallery - 3 images

MIT's robotic cheetah has gained a new sidekick, and the nimble sibling of the four-legged galloping droid has a few impressive tricks of its own. Smaller, lighter and not to be outshone, the Mini Cheetah's acrobatics have been showcased in a video shared by the researchers, including some very natural looking movements and a gracefully executed backflip.

The Mini Cheetah is far smaller than the latest version of the Cheetah robot, which tipped the scales at 90 lb (41 kg) making it the size of a fully developed large dog. A decidedly nimble 20 lb (9 kg), the Mini Cheetah can trot along untethered just like its predecessor to cover the ground at 2.45 meters per second, but its smaller frame allows it some impressive agility, too.

This includes an ability to spin around on the spot, land upright when dropped from height, maintain its balance when kicked by a human and pick itself up again when pushed over. But most impressively, it can perform a perfect backflip from a standing start on flat ground.

The MIT Mini Cheetah robot can pick itself up when knocked down
The MIT Mini Cheetah robot can pick itself up when knocked down

This calls to mind the gymnastics abilities of Boston Dynamic's Atlas humanoid robot, which first impressed with a graceful backflip and clean landing back in 2017. But according to MIT researchers, their Mini Cheetah is the first four-legged robot to perform such a feat.

With its eerily natural running style and ability to get back up when knocked down, it's not the first quadruped robot to evoke haunting images of a Black Mirror episode called "Metalhead," in which a robotic dog wreaks terror over human protagonists in a post-apocalyptic world. The SpotMini from Boston Dynamics also features unnervingly natural mechanics and recently learned how to open up closed doors, a disconcerting sign for those fearful of a robot apocalypse.

To see the new Mini Cheetah do its thing, check out the video below.

Source: MIT (Youtube)

Backflipping MIT Mini Cheetah

View gallery - 3 images
5 comments
5 comments
Daishi
Boston Dynamics started as an MIT spinoff in 1992. They have gone almost 30 years without releasing a product surviving on mostly grants. The Honda ASIMO started with the P series in 1993. The Cheetah mini is really impressive work for something completed by only 3 students but it's academic. Despite some significant mobility improvements 30 years of work by some of the brightest minds in the world has not been able to produce a legged robot that solves a real world problem.
Rustin Lee Haase
The scale may be wrong, but technology like this could make for a fantastic alternative to riding around on a motorcycle or in a car. If only I could ride on top of one of these like a horse. Throw in even more modern engineering and almost religeous zeal of people like the Tesla Motors group and you might just find yourself riding on top of something like this at extreme speeds. (Like David the Gnome riding on top of his fox friend "Swift") No need for roads even, just good solid places to plant running feet. Perhaps the robo-fox could chomp down and swallow you so you could ride along safely inside. The analogy probably should stop there, considering the matter of getting out. :-)
Daishi
@Rustin Lee Haase That's kind of the problem. Wheels are inherently several times more efficient than legs. Compare a human walking vs a human on a bicycle for instance. Legged robots are Rube Goldberg machines with no logical use. Building something like this is a useful learning experiment for engineering students but government grants should not fund legged robots for some of the same reason it should not fund homeopathy, astrology, or perpetual energy machines. It matters because there are other more viable mobility platforms not being funded or researched.
Sharky67
Projects like this need not be discouraged because of the (accurate) assumption we're not progressing such devices through to common-place adoption or role-based service. At the very least, if they're an educator that furthers science/engineering/design toward better outcomes, then such programs need to remain. However, they benefit the younger more pliable and curious mind, and when it comes to grants or funding, that target needs to remain in focus. Not some corporate entity whose profits will never be granted back into education.
GregVoevodsky
Legs have a place in this world that beat wheels. When the road ends and climbing begins like Everest or trails in the Grand Canyon or Aphganastan - a good mule or 4 footed or 2 footed robot can do more. How about a 9/11 footed vs a tank tread? It's good to develop both. It's funny how nature never invented the wheel probably because wheeled animals would get stuck and die vs footed or winged. ;-)