Two years ago a cheetah-inspired robot “ran” at 28.3 mph, faster than Usain Bolt. A new rival model can only manage 10 mph, but does so without any external power cables.

Both Cheetah model are funded by the American military’s gadget department DARPA (Defense Advanced Research Projects Agency) but are produced by different firms. In 2012, a Boston Dynamics Model achieved a peak speed of 28.3 mph on a treadmill.

That made the headlines as that speed was taken across a 20 meter run. During his world 100 meter record-setting race, Bolt averaged 27.78 mph in his fastest 20 meter section. (To be fair, there are several caveats to that comparison, most notably that being on a treadmill rather than a track was calculated as advantage equivalent to a tail wind six times that allowed when setting athletics records.)

Now the Massachusetts Institute of Technology has announced that its Cheetah model can run entirely on battery power and thus isn’t restricted to a treadmill. Carrying the weight of a battery pack slows it down to around the 10mph mark, but it can maintain that speed for more than 15 minutes.

Sadly that means it wouldn’t win any athletics races. It would be handily beaten at all distances up to 3,000 meters and would run out of power with a couple of laps to go on the 5,000 meters.

It’s still an impressive performance however, and as an added skill the robot can leap up to 16 inches high during its run. Its frame, 12 motors, on-board computer and batteries add up to 70 pounds in weight.

The biggest design challenge was producing the four legs using a material strong enough to support both the weight of the robot’s body and the impact of running and landing. That was achieved through a combination of Kevlar strips in the legs and a frame that sandwiches foam between carbon fibers.

At the moment the robot, which may be used for search-and-rescue missions including in hostile territory, is entirely remotely controlled. One of the next development steps will be adding sensors to allow it to determine its own course and movements.

(Image credit: MIT)