This is a drive train concept I have been thinking about now for a little while and I’m just getting around to prototyping it with VEX. In the images below is one of the nearly completed first revisions of a drive pod.
Feel free to post comments or ask questions and I’ll continue to update this thread as this experiment continues. At the very least I think this will be a cool test bed for the programmers to use.
So that’s the main problem - very hard to do this with VEX since they don’t have 45 degree bevel gears (current bevel gears are intended for 90 degree motion transfer) I looked at using just one of the bevel gears into a spur gear, but that connection was messy and probably would have destroyed itself very quickly.
So what I am currently doing for Rev 1 is powering all of the holonomic wheels independently and having the traction wheels linked together with HS chain and sprocket. Each of the drive side’s pod’s traction wheels will be chained to one another, requiring only one additional motor per drive side for each of those wheels (we’ll see how that works too).
The other part of this that people will probably ask about is how the drive pods intend on pivoting. Ideally we’d use the VEX pneumatics kit as it’d be perfect for this application. However, we don’t have a VEX pneumatics kit.
At this point, I’m looking at using a short linear slide with a linkage connected to each sides’ drive pods to control which wheel is on the ground. Not as cool as pneumatics but it will work.
Nice concept, but I fear getting around that corner isn’t going to be easy. You could use a bevel set to change the axis of rotation, chain past the 45 then use another bevel set but that gets costly in both size and complexity.
I do like it though, all good in the name of attempting to make a better holonomic drive.
If college team, custom plastic 45 bevel gears would be possible.
Existing bevel gears are reported to be low torque, so consider N:1 speedup, bevel gear, reduction to final drive wheel.
Other alternatives for 45 angle drive:
Custom machined CV joint
flexible link as U joint
Tilt-wheel-subframe activation requires lifting the entire robot weight.
Servos seem unlikely to be strong enough.
Alternative actuator might be a worm gear on a linear track.
Other alternatives along similar lines:
GreenEggs type Mechanum wheels, to avoid the 45 angle, more like octocanum on CD.
Servo/pneumatic activated side-clamp-plate to lock the rollers on the omniwheel.
Use + holonomic instead of x holonomic, and alternate with tank drive wheels.
My thought on servos is that they are going to cause you big problems. You have to lift the entire robot up when you switch between drive wheels so you are going to need a good amount of torque and on top of that you are going to need to hold that position as you drive in your traction setting which will likely cause the servo to heat up and not work as nicely as you would like it to.
I would go with pneumatics. They are strong and they have no problems holding their position, plus it frees up some motors/servos for use on other mechanisms. The pneumatics aren’t cheep but i think they are vital to your design.
The worm gear on a linear track was exactly the idea I was thinking of using, one for each drive side with linkages that attach to the drive pods. The amount that the pod must pivot to switch between wheels is not a large angle, and the worm gear set would help prevent back driving when one set of wheels is on the floor.
Agree with your sentiments on servos. Unfortunately pneumatics are not an option here. It’d be great if they were but that’s not the case. I don’t think they are ‘vital’ to the design, they’re just probably the most simple solution, but they’re definitely not the only way to actuate the drive pods.
I should also add that the weight of this currently is pretty substantial being that I’m just using some of the extra steel we have for this first revision. Look for a lot more polycarb in the next revision, if it gets that far.
Wow! That is a really neat idea. I like the image you posted. Maybe you can make yourself a universal joint of sorts to get around that angle? Just a thought.
I think that FRC team 148 has made something similar to this. Their drive has 4 high traction wheels and it has a pivot like you discussed that switches to 4 omni wheels. However, instead of mounting the wheels at a 45 degree angle, they deploy a perpendicular wheel when the drive pods pivot. I think that they pivot their drive pods using pneumatics too. Skip to about 1:10 in this video. Robowranglers 2010: Armadillo - YouTube
EDIT: Sorry, didn’t read your first post too carefully there =/
Correct, as I stated in my original post the Robowranglers were one of the first teams I saw to do this. Their 2010 and 2011 drives heavily inspired this concept.
One of the nice things about the HoloTraction drive concept would be the elimination of the set of perpendicular omni wheels required in H-drives. The drive pods take up more room on the outside of the frame though so you’re losing some of that space, but gaining more open space in the center of the frame, which is always nice for mechanism placement.
The drive portion for Rev 1 is complete. An actuator driven by a worm gear was prototyped and worked well without back driving, although it was much slower than desired. We’re also looking at actuating all four drive pods with just one motor.
Note: One of the HS motors is flipped for some reason… need to fix that.
This was actually something I hadn’t really thought too much about and ran into when assembling the first drive side. However, the neat thing is that when the pods are in traction mode and are pushed down more on the traction wheel side of the pods, the chain actually gets pulled tighter as the sprockets actually move farther away from each other.
So the chain is too loose when in holonomic mode, but is perfectly tensioned when the drive is in traction mode. I’ll take a few more pictures of this on Tuesday.
I would still suggest pivoting on the traction wheel axle. Even if it’s not used in holonomic mode, loose chain should be avoided…I would be worried about the chain jumping in strange ways as it gets “re-tensioned.”
Exactly if you look at the nonatread it pivots about the axle its driven. The student wanted. To do exactly what you are doing here but pivoting about the axis just reduceces the chances of failure. I would strongly reccomend redesigning to account for this.
Thanks for the suggestions, we will definitely keep it in mind for Rev 2 (or maybe Rev 1.5, it wouldn’t be that hard to have the pods pivot on the traction wheel’s shaft instead). This is why I like posting the concept publicly, constructive feedback is always appreciated.