Sphere bots get some new skills

posted Oct 15th 2009 1:00pm by Caleb Kraft
filed under: robots hacks

Spherical robots , or in this case RC vehicles are pretty versatile. They travel about the same on most terrains, including water in some cases. That’s not to say that they travel particularly well on those terrains though. The common problem is that they can’t really climb over bumps very well, until now. We’ve seen a few versions of sphere bots, but they all seem to need fairly level smooth surfaces, aside from that one that went in the water. We hadn’t seen any that really had the oomph necessary to climb stairs though. Actually, we still haven’t seen that, but he says it can in the interview you can watch after the break.

Comments [34]

tagged: sphere

Reader Comments

I made a sphere bot using a hamster ball and a series of motors with weights on them. They are fast and fun. Would love to see how this thing conquers stairs.

Posted at 1:03 pm on Oct 15th, 2009 by Pard

not gonna lie, i’m way more interested in the tank

Posted at 1:08 pm on Oct 15th, 2009 by cyc4015

anyone catch the obligatory automated dorm room door at 1:45 lol

Posted at 1:59 pm on Oct 15th, 2009 by twistedsymphony

@ cyc

2nded

Posted at 2:01 pm on Oct 15th, 2009 by McSquid

the tank looks awesome.
In one of our previous posts we showed a tiny bit of a radio controlled sphere I hacked together with RC car parts. It was sloppy, but worked. It would barely go over anything though.

Posted at 2:04 pm on Oct 15th, 2009 by Caleb Kraft

Oooh, a tank!

Posted at 2:08 pm on Oct 15th, 2009 by mars

also, i imagine what makes the ball manage obstacles so well is that it weighs a freaking tonne

Posted at 2:17 pm on Oct 15th, 2009 by cyc4015

appears as though they used something (a pen?) tucked under the rubber for better traction. would a final design include many of these ribs around the sphere?

Posted at 2:32 pm on Oct 15th, 2009 by lackofmotive

@cyc4015
3rded

Posted at 3:14 pm on Oct 15th, 2009 by monkeyslayer56

cyc4015: Nah they seemed to be saying it uses gyroscopes to provide a kind of solid base to push off. I guess you could also use a momentum wheel in the same plane as the axle if you wanted. That might actually be easier.

Posted at 3:59 pm on Oct 15th, 2009 by Tim

It uses gyroscopes to navigate the terrain so well.

Posted at 4:13 pm on Oct 15th, 2009 by Dan Cardin

I’m with the others – screw the ball, I want the tank! Oh yeah, and I’ll take one of those pressurised air-cannons mounted on it too ;)

Posted at 4:14 pm on Oct 15th, 2009 by Daley

@twistedsymphony: i live in the dorm at MIT where that automated door is! you can actually open/close it with an iPhone app. :)

Posted at 5:13 pm on Oct 15th, 2009 by Steohan

i basically want to be an inventor too, and have the resources to just tinker.

Posted at 6:02 pm on Oct 15th, 2009 by kuhl

Agree with others previous comments – it’s easy to make a sphemical bot with plain motors and drive it around. You can never get great steering, or more importantly stopping control with these designs though

This implementation really is excellent – the gyro seems to make it much more stable and controllable.

Posted at 6:24 pm on Oct 15th, 2009 by trialex

@kuhl: i’m sure a lot of readers hear have the same goal

Posted at 6:29 pm on Oct 15th, 2009 by john

That’s me in the video, I’m glad you guys like it. I’d be happy to answer questions about it (or tanks or cannons or dorm room door actuators).

Posted at 6:56 pm on Oct 15th, 2009 by Greg

@Greg: real nice !

Is that the hydraulic door opener that uses the building water supply as a source? Great out-of-the-box thinking :D

How do the gyroscopes help? Are you using two like this [http://blog.makezine.com/archive/2009/09/how-to_use_hard_drives_for_image_st.html] to add ‘inertia’ to the pendulum part? Or just electronic ones to do automatic closed-loop stabilisation?

Posted at 11:27 pm on Oct 15th, 2009 by Jonathan

It certainly is. Making it hydraulic was more fun than just slapping an electric linear actuator on there. And it has the added benefit of being manually overridden and incredibly quiet in operation.

The gyroscopes work by storing huge amounts of angular momentum. They are used in way that makes them whats called control moment gyroscopes (http://en.wikipedia.org/wiki/Control_Moment_Gyroscope). They arent used for stability like in that example with harddrives. Their stored angular momentum is dispensed to create extra torque.

Posted at 11:46 pm on Oct 15th, 2009 by Greg

Nice idea)

Posted at 1:43 am on Oct 16th, 2009 by sunny185

Two words:

AWESOME

Posted at 2:07 am on Oct 16th, 2009 by _matt

So your robot still changing the gravity center to move arround, but when is stuck in a hole or something similar you use the gyroscope to create extra torque and get out from there no?

By the way, awesome robot where do you get the sphere?

Posted at 2:10 am on Oct 16th, 2009 by Gizer20

@Greg: indeed, hydraulics rock :)

So I suppose (also from the pictures) there are two gyros spinning in the horizontal plane, I think you are only using them as a angular momentum storage, and then simply driving the ball against this momentum-heavy pendulum. (since attempting to tilt them will cause them to react by tilting left to right in opposite directions, which they cannot)

But how about simply attaching the gyros to the inside of the sphere on left to right gimbals, without using a pendulum, and then tilting them?
This should create forward torque I think, and would sound more like a CMG to me, no?
Not sure about turning though…

Posted at 2:13 am on Oct 16th, 2009 by Jonathan

Hack a segway and use its guidance system to buffer out all the wacky movements to give better control?

Posted at 7:05 am on Oct 16th, 2009 by Mic

@Gizer20: Yes the robot still shifts center of mass for normal motion. The stored momentum gives the pendulum more than just gravity to push against to create torque. It uses both at the same time.

The shell is actually a polycarbonate gumball machine globe.

Posted at 9:08 am on Oct 16th, 2009 by Greg

@Jonathan: The gyros are actively driven to rotate left/right to generate the precession torque. So they are in fact used like CMGs.

The reason you can’t just use CMGs is because you can only store so much momentum. By conservation of angular momentum, you can only generate a certain amount of torque*time before your CMG system is ‘depleted’.

Steering is accomplished using the pendulum, but can be supplemented by the CMGs as well (in a slightly different/more complicated configuration).

Posted at 9:24 am on Oct 16th, 2009 by Greg

@Mic: The control system for this robot is significantly more complex than for the segway. I’m building a custom system with an IMU and encoders on all the parts. I’m also developing the dynamic equations of motion mathematically, which are all delightfully non-linear. After that I’ll be able to write the software to cancel out all the wacky movements.

Posted at 9:29 am on Oct 16th, 2009 by Greg

This is ingenious. Nice work!

Posted at 10:04 am on Oct 16th, 2009 by Andrew

@Greg: how is it that the gyro you have on the table is staying up? Is the lighting in the video obscuring something that I should be seeing or is it just a metal stick with the spinning ball parts on the end…and standing up on its own?

Posted at 5:55 pm on Oct 16th, 2009 by Dan Cardin

Greg you’re absolutely right. What I meant to convey was the idea of compensation. You would obviously need a more complicated and specialized system, but my proposed concept has merit. Something designed to measure and compensate for g forces. I mean you’re right it is more complex, but you still need a computer to measure forces and to compensate for the ones that mess up your control inputs. That’s what I tried to say, but, a segway is a lame example as you have said =P Awesome creation, serious props man.

Posted at 7:52 pm on Oct 16th, 2009 by Mic

@ Dan Cardin

the stick with the ball has a gyro in the ball. when the gyro is spinning, it keeps itself positioned in relation to its original position and resists changes in yaw thereafter. once the gyro stops, the whole assembly falls.

Posted at 4:16 pm on Oct 17th, 2009 by Madmaxx

Well, thats nifty then, since it seems to defy gravity.

Posted at 8:05 pm on Oct 17th, 2009 by Dan Cardin

http://www.youtube.com/watch?v=4KlddP16uuU

theres more videos by gyroscopes.org that demonstrate some pretty cool gyroscopic precession.

Posted at 8:35 pm on Oct 19th, 2009 by Greg

i made a sphere bot.. it had a hamster ball with a hamster inside. it rolled away and never came back :(

Posted at 4:34 pm on Oct 26th, 2009 by none