Crazy Engineering: RoboSimian Robot

Hey guys, imagine a disaster situation that's so dangerous we can't send people in to fix the problem. So what do you do about that? Well, this is RoboSimian and that's exactly what he's designed to do. We're going to talk all about that on this episode of Crazy Engineering.

(Music)

Alright guys we're here with Brett Kennedy, he's leading the charge on the RoboSimian project.

Hey Brett, could you just tell us a little about RoboSimian?

What were you guys trying to accomplish with the robot.

RoboSimian is actually a line of limbed robots that we've done here at JPL, but specifically designed and built for the DARPA Robotics Challenge.

It's disaster response, not life saving.

So the example that DARPA uses is the Fukushima nuclear reactor. After the earthquake and tsunami, there was a lot of damage to the reactor and they really needed to get inside and do a few things that would've made everything better. And

it could be as simple as flipping a switch or turning a valve.

So JPL is kind of known for making rovers and everyone knows that rovers have wheels and they climb over rocks. Why did your team decide to go with limbs instead of wheels?

Now if you can get away with wheels, wheels are great. They're the most efficient ways of getting around. But once you get into terrain that's too rough, then we need to switch over to a different method.

So how does the robot know where it is and where it's going?

So if it can it's going to use its perception system, its cameras and lidar, to actually build this 3 D map and it sends that information back to the operator. The operator then decides where it should go.

But even it doesn't have good data about what its environment looks like, it also has got 4 sensors on its wrists and ankles so it can actually feel the terrain as it is walking as well so we have the capability of basically moving around by braille.

So what exactly can it do?

Given the fact that it's sort of human scale, we can have the robot do a lot of things that a human can do, actually, as long as you're not too concerned about it moving fast.

What kinds of hands or endofactors can you put at the ends of these limbs?

We went though a couple sets of hands and this is where we've ended up today.

It's not for dexterous manipulation of things but it can do most things you need, which is to grab on to objects, being able to pull with a lot of force able and also be to manipulate certain things, in particular human tools.

So looking at this arm it looks pretty advanced actually.

This limb is actually very similar to yours in other ways, particularly the number of joints it has.

So even though it doesn't look like it, its about as dexterous as your own arm.

You've got seven joints, I'm assuming seven different independent motors? Tell us a little about these motors. What torque capabilities they have? How fast they can go?

In every one of these joints we have exactly the same drive train. This contains electronics that run it, the brake, the motor itself, and the drive train, which is a harmonic gearing in this case.

What's your gear ratio out of that harmonic drive? 160 to 1.

That's a pretty tight package - really high gear ratio.

Yeah. Out of this one little thing we can get as much torque as a F150 truck.

So I'm imagining RoboSimian kind of crawling around buildings and over difficult terrain and that's all happening on Earth. Is it even possible to send this to space?

Well the exact technology that goes into that would be different from what we have in RoboSimian today but the basic robotics problem is the same. So yes, absolutely we could send these to the cliffs of Mars, to the outside of asteroids and comets, things of that nature.

Alright, we're all wishing the best for the RoboSimian in the DARPA Robotics Challenge this June. Good luck out there buddy!