Slashdot Mirror
Novel Drive Wheel System Based On Spinning Sphere
An anonymous reader writes "A Bradley University student has built a mobile robot that uses a hemispherical omnidirectional gimbaled, or HOG, drive wheel. It consists of a black rubber hemisphere that rotates like a spinning top, with servos that can tilt it left and right and forwards and backwards. The HOG system delivers an amount of torque directly proportional to the tilt of the hemisphere, allowing the robot to move incredibly fast nearly instantaneously."
So, the funky cars they used in the iRobot movie with Will Smith, this re-invention now makes them possible? I am curious to know how well it works on rougher surfaces, like potholes, sand, or gravel.
It seems interesting but I have a few concerns (some have been stated in other posts but I would like to get them in one place).
1. Uneven surfaces; With such a small surface contact it is easy to lose traction.
2. Control. It seems that one can change direction at will but it seems difficult to do it accurately and more difficult to stop the device.
3. Soft surfaces; if the hemisphere is going at a constant speed would it dig into a sift surface when stopped? Sure you can stop the motor but that means you would have the same acceleration characteristics of a conventional wheel.
I would have liked to see it on bare concrete doing a slalom and stopping at a designated point.
Actually, I think this would be good for robots as they could adjust their parameters to counteract the bumps and stuff really quickly. Maybe a drive-by-wire system on a car where the computer does all the hard stuff and you just point the car where you want it to go.
I could really see this being used in high speed robotics applications, imagine a ground-based sentry drone with this screaming down the road.
Job? I don't have time to get a job! Who will sit around and bitch about being broke and unemployed then?
I don't get all this worry about accuracy and precision. Wheel/tyre wear and surface undulations only matter if you need to position this robot by dead reckoning. Dead reckoning is a bad idea anyway except in machine tool or laboratory applications.
Get real. This type of drive system would be for applications like vacuuming floors and moving stuff in warehouses. Its positioning would be determined by external feedback, like lines on the floor in a warehouse or ultrasonic echoes from walls. Other factors would be irrelevant to positioning.
As for the guys worrying about loss of contact and friction (someone raised "omni-wheel" designs, with wheels composed of little wheels arranged around a big wheel) what is the problem? In TFA' photo I see a three wheeler (one HOG wheel and two conventional idlers - though it could be developed to three HOG wheels only). How can a three-wheeler lose tyre contact ?
It's a cute idea. It assumes a single point of contact with the ground, and thus requires a flat, hard floor. This is limiting.
I've worked pretty extensively with mechanum wheels - essentially omniwheels with the smaller wheels at a 45 degree angle to the main wheel. Arranging four of them provides the same degrees of freedom as the example shown with two of these HOG wheels. Mechanum wheels work well and move quite fast, and I've yet to see a surface where they don't work - but they're costly, heavy, and wear quickly, not to mention the pretty enormous power requirements. Because of these limitations, for hobbyist robotics, they're simply not practical.
For many of the smaller projects I've done, traditional drive systems were slow and not nearly as useful as an omnidirectional (3 DOF) system - and without the ability to easily use something like omniwheels or mechanum wheels due to various constraints, HOG wheels would be a godsend. They provide most of the benefits of the traditional omnidirectional drive systems with very few hitches - and you'd be surprised how often the hard and flat surface requirement isn't an issue (or, in many cases, applies to traditional drive systems as well).
Yes, I understand the physics, but I was really countering some guys here who seem to think the wheel contact requires laboratory precision and are worrying about slight wear (needing computer compensation!!) and loss of contact (how???).
I agree this drive system is never going to be an off-roader. Also, many posters here seem to be floating the idea of it being used for cars - forget it. But as I have said in aother post here, the ideal use for this drive is fork-lift truck type applications, where tight manoevering is required, and, incidentally, the floor is flat.
But I don't think it will be as sensitive to undulations and kerbs as you and others seem to think. Most "bumps" in the floor/road are actually at quite small angles unless they are actual holes with rims. Also, a spinning top reacts violently (eg shoots across the room) when it hits the wall because its whole mass is spinning, so it has a lot of rotational momentum to be converted into linear momentum. However, this HOG vehicle has only a small proportion of its mass spinning, so contact with a kerb will probably just result in a nasty jolt and some rubber left behind. For warehouse use, just put guards around the wheels (as the prototype has).
You are worried about cracks?! I think we are at cross purposes about the scale of this thing. I assume that the prototype is a model, but even then I do not know its scale. For real world use (eg the warehouse vehicle) I envisage HOG wheels being hemisperes of about 400mm diameter. You would have to find an extremely run-down warehouse to have cracks in the floor that big!