Hemispherical Omnidirectional Gimbaled Drive

[SenorZ]

Just saw this Gizmodo.com and found it most interesting.
Youtube video: http://youtu.be/uaT7M3Nwj7c

I'd love to see someone scale this drive system up to FRC size. The popularity of swerve drive, and other omnidirectional drives, isn't going away anytime soon. It seems that a major technical hurdle facing teams is building the wheel units and linking them all together... and having multiple transmissions.

The maneuverability looks promising.


Oh, yeah, is this the correct forum for this topic? I don't post much

[Cory]

http://www.chiefdelphi.com/media/photos/15197

Similar, though the angle of the hemisphere with respect to the ground is static.

[Lil' Lavery]

There's a pretty significant difference between a ball drive and what this does. This doesn't have to change the speed of the motor(s) to change direction, only the angle of the "wheel." The ball drive has to decelerate and accelerate the motors controlling the balls in order change directions.

[PAR_WIG1350]

cool, but one of the issues I see with scaling this up to frc size is that each drive unit would require 2 steering motors, on the other hand, a common shaft could be run through the robot from which all wheels are powered through chains and universal or cv joints. But still,the minimum number of motors for a robot supported entirely on its driving wheels is seven if three wheel units fed from a single motor driving a common shaft. if only two driving units are used, that drops down to five motors, but you now must support part of the robot on non-powered wheels which reduces available traction.

The greatest disadvantage would be that both the seven and five motor versions that I mentioned use only one motor for actually driving. in comparison, a six motor swerve drive configuration uses three of the six motors to drive and a four motor meccanum system uses all four motors to drive, but at reduced efficiency due to canceling vectors.

The final concern would be wear on the field and on the hemispherical wheels, the sliding motion that occurs when the wheels pivot on the gimbals would result in a lot of wear on the tread. Also, when a wheel is idling (when its axle is vertical and it is not contributing to motion) that part of the hemisphere, not to mention the carpet under, could experience accelerated aging and wear due to the friction and heating that would occur. The solution to this problem would be to use a low friction material like acetal (2009 wheel tread material) form the hemispheres' surfaces, but then you would have little to no traction. [In the video, he is driving on a slick floor which reduces this issue].

if these issues could be dealt with (if a team can devote 9 motors to drive and implement a weight transfer system to reduce high-wear interactions with the floor by unloading drive wheels while idling (pivoting could still pose issues, just have spare wheels to replace worn ones I guess), then it could be done and would likely win an award (specifically innovation in control). But the question is, who is willing to face the challenge*?






*probably not us, we tried an 'innovative' {read "complicated"} drive train in 2010, it wasn't pretty. On the other hand, watching a robot drive across the field as parts of the drive system fall out behind it leaving a trail of parts while it continues to drive is rather comical.

[Colin P]

seems like it would've been good in 08'

[davidthefat]

It seems that it would be hard to scale up. Scaling up to FRC level would be fine, but to a commercial level would not be all that easy. What I imagine is that the wheel platform would just snap. It's like trying to stand on one leg and trying to rotate your foot. I think there would be too much torque acting on that to be safe.

[Lil' Lavery]

Originally Posted by davidthefat

It seems that it would be hard to scale up. Scaling up to FRC level would be fine, but to a commercial level would not be all that easy. What I imagine is that the wheel platform would just snap. It's like trying to stand on one leg and trying to rotate your foot. I think there would be too much torque acting on that to be safe.

Do you have any calculations, even first order estimations, to support this hypothesis? And since when was "commercial level" automatically larger than FRC scale?

[s_forbes]

Great link, thanks for posting! It's a very simple method, construction wise, of creating a drive module capable of moving in any direction.

Originally Posted by PAR_WIG1350

cool, but one of the issues I see with scaling this up to frc size is that each drive unit would require 2 steering motors, on the other hand, a common shaft could be run through the robot from which all wheels are powered through chains and universal or cv joints. But still,the minimum number of motors for a robot supported entirely on its driving wheels is seven if three wheel units fed from a single motor driving a common shaft. if only two driving units are used, that drops down to five motors, but you now must support part of the robot on non-powered wheels which reduces available traction.

The greatest disadvantage would be that both the seven and five motor versions that I mentioned use only one motor for actually driving. in comparison, a six motor swerve drive configuration uses three of the six motors to drive and a four motor meccanum system uses all four motors to drive, but at reduced efficiency due to canceling vectors.

Just because the drive modules used in the video each use two actuators doesn't mean you'd have to implement it like that on a FRC robot.

For some systems it's easier to think about it in terms of degrees of freedom (DOF); for example, the typical "full swerve" drivetrain is able to move in x, move in y, and rotate... a total of 3 DOF. 3 DOF can ideally be obtained with just three actuators.

As an example, the Robonauts have effectively used a full swerve drive system in the past by using a crab drive (all modules steer together with one motor, all modules are driven together with one motor) with a rotating 'top' manipulator to give rotation. 3 motor inputs = 3 DOF.

Along that same logic, a typical six wheel drive train has two degrees of freedom: rotate, and drive forward/back. This requires two inputs, left side of the drivetrain and right side of the drivetrain. 2 motor inputs = 2 DOF.

The module shown in the video is intriguing because the tilting of the module acts somewhat as a gearing system. The further it tilts, the faster you can go. This effectively gives you a high gear at full tilt, and low gear with minimal tilt, as well as every gear in between (CVT!). So, tilt in x and tilt in y = 2 motor inputs = 2 DOF for a single module, the outputs being velocity and direction of travel.

Practically, I don't see a good use for this kind of system for an FRC robot. A simple swerve module already gives directional and velocity control (via motor speed controller). Any advantage that this system has in terms of variable gearing is most likely lost in the hemisphere/carpet interface... I can't envision getting a respectable amount of traction in low gear with such a system. The fact that it effectively works as a continuously variable transmission is appealing, however. I'd like to see one made, in Robonaut fashion! 4 modules tilting together with a central power plant to drive the hemispheres... 3 DOF for a CVT crab drive.

[Andrew Remmers]

So after seeing this drive, I questioned weather it was possible at a VEX scale.

Sooo I went and spent a few hours CADing and went through two iterations of gimbles.

This is the final version

Comments and critiques are appreciated

Thought everyone would like this!

- Andrew

[Ankit S.]

Originally Posted by Joyride_67_1902

So after seeing this drive, I questioned weather it was possible at a VEX scale.

Sooo I went and spent a few hours CADing and went through two iterations of gimbles.

This is the final version


Comments and critiques are appreciated

Thought everyone would like this!

- Andrew

That actually looks very awesome! What are you going to to make the hemisphere out of/ Where are you going to get it?


On a side note, we should call this the HOG Drive (Hemispherical Omnidirectional Gimbaled Drive)

[Andrew Remmers]

Originally Posted by BeltSanderRocks

That actually looks very awesome! What are you going to to make the hemisphere out of/ Where are you going to get it?


On a side note, we should call this the HOG Drive (Hemispherical Omnidirectional Gimbaled Drive)

Is this in reference to Aperture's "Hand of God" (HoG) Or Exploding Bacon?

Just has to clarify

The material we don't know what it will be yet since this will be made for College Level VEX if we make it (prototyping tomorrow)

- Andrew

[Ankit S.]

Originally Posted by Joyride_67_1902

Is this in reference to Aperture's "Hand of God" (HoG) Or Exploding Bacon?

Just has to clarify

The material we don't know what it will be yet since this will be made for College Level VEX if we make it (prototyping tomorrow)

- Andrew

I made that up right now

If you are indeed prototyping it tomorrow, video please?

[Andrew Remmers]

Originally Posted by BeltSanderRocks

I made that up right now

If you are indeed prototyping it tomorrow, video please?

To Funny!

We will see how far we get the only thing we probably won't get the hemispheres made we will probably try to make a makeshift version of the hemispheres. Pictures and video will be made!

-Andrew

[davidthefat]

Would a lone drive hemisphere drive train mounted in the middle of a traditional 4 wheel omnidrive system work? The omni wheels would not be powered, but just used as castor wheels.


And Joyride, great work. I am impressed. Only critique is that it looks very back heavy, or is that a fundamental part of the drive system?

[Andrew Remmers]

Originally Posted by davidthefat

Would a lone drive hemisphere drive train mounted in the middle of a traditional 4 wheel omnidrive system work? The omni wheels would not be powered, but just used as castor wheels.


And Joyride, great work. I am impressed. Only critique is that it looks very back heavy, or is that a fundamental part of the drive system?

Yes a lone hemisphere would give you the same effect I guess I personally don't like the idea of mounting a ton of omniwheels everywhere since the wheel on the gimbal changes the ground clearance making the robot need at least a form of suspension to keep all omniwheels on the ground, at least thats what I would do.

Also may I point out the Omniwheels and structure up front was for a ground test to see what angle the drive would sit when the robot was essentially in "neutral"