Cheetah, a robot being developed by Boston Dynamics under the DARPA Maximum Mobility and Manipulation (M3) program, is making the rounds on the Internet again, as it has reached speeds of 28.3 MPH on a treadmill. This thing is crazy impressive:
I know a few of the guys over at Boston Dynamics working on Big Dog/AlphaDog and PETMAN/Atlas, but I’ve never really met anyone working on Cheetah. It’s an awesome project, worthy of every bit of jealousy that I’m feeling right now. One that I like to follow, because even though we often work with Boston Dynamics at IHMC we’re also kind of “competing” against them. We’re developing our own fast running robot, that we’re calling FastRunner, which aims to achieve the same goals using wildly different approaches.
Those of you who were followers of me back on Tumblr might remember last year when I wrote about FastRunner; I lashed out at Gizmodo for publishing an under-researched piece that gave all of the credit for our robot to MIT and DARPA. I’m still a little chafed about that, but then again I’m always mostly chafed at Gawker.
Regardless, the lab is gearing up to start putting the pieces of FastRunner together so that we can have a working prototype before the end of October. Like Cheetah, FastRunner aims to reach incredible speeds on land by utilizing hydraulics as a power source and biologically inspired mechanical design to enforce robust and dynamic systems. That’s where the similarities end, though.
FastRunner was dreamed up as a very novel concept; instead of more “traditional” designs where all of the joints in the body are motorized and the stability is reenforced by on-board computer control systems, FastRunner aims to achieve its goal by using the natural shape of its design to remain stable, upright, and fast. It’s modeled after an ostrich, and by utilizing a mechanically impressive array of springs and locks to mimic a tendon network, it can transfer a great amount of power and energy using only two powered elements; one actuator at each hip joint. Nothing in the knees or ankles required.
We’ve had it going as fast as 30+ MPH in our simulations, which will obviously not translate directly to the real world, but you can see demonstrations of slower running and the machinations behind an earlier prototype of the legs in these videos:
I, however, am not doing much work on the FastRunner project myself; you see, the requirement for little-to-no onboard control systems means that programmers don’t get to have much fun in this world. And I know shit all about mechanical engineering. But I’ll probably post a little more about the robot and I’m sure it’ll end up on the Internet somewhere once we get it up and running (hah!)
In the meantime, here are some renders from our mechanical design team of what the prototype is going to look like, and you can see movies of our simulations running on our group’s site linked above.