Four Wheel, all steering, non-omniwheel, ONE power/motor source drive

[FizMan]

As some of you noticed, I've been asking a few questions here and there. And a few of you have asked what's up. Well, I've been doing some brainstorming regarding an all-direction drive system (I think you all have a special term for it; but for the life of me it escapes me).

It seems that the simplest and easiest way is to employ the use of omniwheels setup in a perpendicular fashion. But I'm not keen on the idea of splitting up the motor power of the robot between the two directions. Nor the seemingly lack of traction when compared to some other robots (as you really only have two wheels that will be opposing the force; the other two will just roll with it)

So this is what I came up with for my ideal drive system:

1) All the wheels can steer 360 degrees
2) All the wheels can individually go forwards or backwards while having a single motor set, gearbox, and transmission power ALL the wheels.
3) At least a two speed transmission (though to be honest, I'd really like to have an ULTRA-low gear as well)




1) All the wheels can steer 360 degrees

After thinking about this one, I figured there's no point to having the wheels steer individually; that is, to be able to have them facing different directions. While the thought of having them all opposing each other in order to combat being pushed, I figured it might be better to instead have them turn in the direction of the force and use the motors to push back.

So I came to the conclusion to develop a steering system in which turns ALL of the wheels simultaneously. I concocted one system (that has two variations) that seems to do the trick. I'll probably post another animation of it when I get the chance to model it (probably tomorrow evening), so I'll come back to this later.



2) All the wheels can individually go forwards or backwards while having a single motor set, gearbox, and transmission power ALL the wheels.

This is one of the bigger problems I've been facing. When I was putting such a drive system through the ropes (in my head) it dawned on me that as the wheels turn, in order for the robot to spin, the wheels that turn backwards/forwards change. That is, say the wheels are all pointing forward and you want to turn right, well, simply you spin the left wheels forward, and the right wheels backwards. Voila. So I originally thought to have two drive systems; one for the left wheels, one for the right wheels. BUT, say the wheels are all facing left: now you want the forward wheels to spin backwards, and the rear wheels to spin forwards in order to turn the robot.

So now I'm faced with two choices: a) Build a motor-mount, gear box, and transmission for EACH of the four wheels
or
b) Use a single motor-mount, gear box, and transmission that would power ALL the wheels simultaneously AND develop some sort of method to reverse that power for individual wheels.

And personally, while a) MIGHT be the more feasible choice... to my knowledge, option b) hasn't been explored much. And it might turn out to be a good choice too; this would save grief having to build four gear boxes, four motor mounts, and four transmissions. Plus I like a challenge; I want to see it done



3) At least a two speed transmission (though to be honest, I'd really like to have an ULTRA-low gear as well)
I hope this is pretty self explanatory. Ultimately, in a perfect world, I would like to have a high speed, low-speed/high torque, and an ultra-low-speed/ultra-high-torque gears. I want this drive train to get down and dirty; and if need be, it will drive slow... but it will certainly move whatever is infront of it.




Now, with regards to 1), I do have a question concerning the idea in my head. I can't elaborate on the idea now, but I will tomorrow evening when I get the time to. But just a quick question: I see chains used all the time; but always in a vertical fashion. Is it feasible to have the chains run horizontally? Or will it be very prone to slipping off the sprockets? If so, what can one do to counteract that from happening?



In regards to 2), you can read up on my "reverse gear" thread here: http://www.chiefdelphi.com/forums/sh...ad.php?t=28174



And finally, in regards to 3), I saw the team 222 3 shifter design that has balls pop out of the shaft to engage the proper gear. (http://www.chiefdelphi.com/forums/sh...threadid=26630) That seems like a grand idea; I really like it. But I also read some cautions regarding the design. You think the design is quite alright to use so long as you take the proper precautions (i.e. material selection/lubrication)? Any other three shifters? Maybe I can use that 4 speed automatic shifter Though to be honest; I prefer to have the shifting in the power of the driver, so if I did implement that, I'd program in an override switch. You think it'd be simplist just to use a single small pneumatic to engage a 2nd gear instead? Beauty part of this is I want to have only ONE transmission in the robot; no need to build two transmissions and mount double the parts.



So yeah; I'm a newcomer to mechanical engineering, so I'm looking for your input; and so far it's been great! And I have one problem that seems to be just out of my reach right now. And that is transferring the power of the motors (from a fixed shaft... or gear... or chain, or what-have-you) to a 360-degree spinning wheel. I'm sure it's been done before, but I haven't seen any threads (though admitadly, I didn't look through ALL of them... but a half decent number none-the-less) regarding that issue.

So thanks again on any input and advice you can toss my way! You've been a great help so far!

[jimfortytwo]

Congratulations, you've just reinvented Swerve Drive. I suspect you have already seen some robots like what you describe in action, and just didn't know it. Actually, its amazing the depth to which you have conceptually predicted designs wrought from tough trial and error. Now that you know its out there, I bet you'll have better lucking finding pictures of such systems on CD and the net. In particular i remember seeing a system using two gearboxes routed out to the four corners with horizontal chain, which sounds close to the design you were moving towards. To my knowlege no one has been using gears to switch directions, though. yet.

To get you started, these videos of 118 in action are what first got me excited about swerve drive:
http://chiefdelphi.com/forums/showth...ight=118+video

[Max Lobovsky]

I have been considering ways to make a drive train like you are describing (and i have thought of the one you suggested) and i came to the conclusion that that method is to complex mechanically. As you can see from the reversing gear box that you put in the post you linked from, the reversing system is quite complex. Basically, i dont think its practical. As jim said, what you described is basically a swerve or crab drive is much more do-able (and has been done).

[Dan Richardson]

My suggestion to you guys would be, look in the White papers, and also considering contacting teams like 111 Wildstang or 79 Captain Krunch. There are others ( such as 118 the robonauts ) but I know both of those teams produce beautiful versions of what you described, in 2 slightly contrasting methods. They also both produce detailed Inventor drawings of their drives. 79 used the " crab/swerve " drive this year, and wildstand only used it in their front 2 modules. But I'm sure that both of these teams would be willing to answer any questions you may have, or direct you to drawings of what made in the past and problems they have run into.

Tho this drive is a bit more complex than others, if built correctly, with others mistakes in mind, it can be quite a beautiful piece of machinery. It is deffinitely not beyond us FIRSTers, because so many teams have produced these style drives before.

[henryBsick]

Ok, I willtry my best.
For the horizontal useage of chains, yes. It is done. Here is a rather complicated pic.here

There is another way. it looks like this M module in the corners drive sprocket in the "v" of the m.
The single chain slant inward for tension to the only driven sprocket. when it turns all four of the modules turn.

What is a module you ask...
Tytus knows best. Here is a pic of the many posted here on CD. Much help and other pics can be found in the beginning section of the 2003 robot showcase picture galleries. http://www.chiefdelphi.com/forums/pi...&quiet=Verbose It is driven throught a powered shaft coming werticly throught the center of the module. It takes a 90 degree turn with two gears ans dives a sprocket which chain drives the sprocket on the wheel. The round disk at the top is where the "bevel gear" is. It is run by the horizontal chain. It is a bevel gear I assume because It has the hole in the middle for the wheel drive shaft to go through. It rotates the whole module. There are not teeth shown in the linked pic.

Ok. now the matter of powering the shaft that goes into the modules. Take from the gearbox, the output shaft. Gear it 1:1 four times, so you knwo have 4 different output shafts. Make a reverse gear like your previous threads inclde for each of these shafts. then, make a braking system for these shafts. Like disk brakes on a car. Now, each of the shafts can run at different speeds in forward and reverse. Then you have to get these shafts to each respective module.

This system in my mind is very overweight due to the 4x braking system and 4x reverse gears. Just an idea. Although it may be crazy. It is a challenge.
It would give you what you ask at the price of weight. Also keep in mind that all four wheels will be running off of one motor, or however many you have in your gearbox, which each have a 40 amp limit.

Wew. Done. Sorry if I am unclear it is late. Ask me about the foggy parts. I hope Tyus will make an apperance in this thread too. He knows whats up and down in the swerve world.

PS
lego systems are also shown in the beginning of the 2003 picture gallery, robot showcase. They also give a good swerve module description.

[sanddrag]

Originally Posted by Henry_222

Here is a rather complicated pic.here

Maybe I'm just missing something but if you have a drive system like in this pic, why would you need reversing gears? What is wrong with reversing the motors? I'm confused.

[FotoPlasma]

Originally Posted by sanddrag

Maybe I'm just missing something but if you have a drive system like in this pic, why would you need reversing gears? What is wrong with reversing the motors? I'm confused.

Best I can figure, it allows you to rotate in place.

We had a swerve drive last year (2003), and we've brainstormed for days on end about the possibilities, so I'd like to think my brother and I could answer some of the questions you might ask.

[henryBsick]

Originally Posted by sanddrag

Maybe I'm just missing something but if you have a drive system like in this pic, why would you need reversing gears? What is wrong with reversing the motors? I'm confused.

That pic just shows the useage of a horizontal chain to turn the modules.

[Rod]

Here is a design we have been working on for 2005. Features: 4 wheel drive, 6 drive motors in one gear box, 2 speed trans, very low CG, 4 wheel steering crab and conv., weight as shown 90 LBS!!

[Rod]

1 more view

[jonathan lall]

Originally Posted by Rod

Here is a design we have been working on for 2005. Features: 4 wheel drive, 6 drive motors in one gear box, 2 speed trans, very low CG, 4 wheel steering crab and conv., weight as shown 90 LBS!!

Disgusting.



In a good way.

[Andy A.]

If the wheels can all turn 360 degrees, then you only need them to turn in one direction. Just point them 180 degrees from 'front' and you've got reverse.

95 acutally produced a drive system something like what I think you mean for the 2003 season. All the wheels were inpendently steered and powered. Drills and Chips for drive, Globes and window motors for steering. The drive motors gearing was through worm gears. I've included a link to some pictures on 95's website. Anything in the picture gallery that is from 2003 may be of some use.

-Andy A.

95's photo page

[Al Skierkiewicz]

Originally Posted by Rod

Here is a design we have been working on for 2005. Features: 4 wheel drive, 6 drive motors in one gear box, 2 speed trans, very low CG, 4 wheel steering crab and conv., weight as shown 90 LBS!!

I hope that the drive train is not detailed in your drawing. There is a need for slip clutches in the differentials to allow turning. Without them the drive motors go into stall during turns. Just two drills or two Chalupas driving the wheels means 200+ amps during turns. Add the two FP, two Chalups and two drills and you have exceeded the current capabilities of the battery and possibly tripped the 120 amp breaker. There was a GM team who had a similar design without the clutches and they repeatedly set the drill motors on fire during turns using one drill for front and one drill for back. This is one application where current sensing is an eye opener. Sorry.

[FizMan]

Thanks for the plethora of responses!

First off Henry; that was exactly what I was thinking for rotating the wheels... in fact exactly; that was the two options I had in mind; the X and the M shape for the chains. Good thing aboot the M is that it uses less chain (right?), but if the chain falls off, then we won't be able to steer. With the X, more chain, but if one falls off, we still have three steering wheels right? Though... I suppose that won't really help. I think I like the M better.

I was thinking aboot it again last night, and I figured that all the wheels wouldn't need to independantly go in reverse. If we wanted to turn the robot, have the "engine" (motors, gearbox, transmission) power the two sets of wheels, just rotate the wheels back to the forward/backward position (a max rotation of 90 degrees) and have them spin in opposite directions.

I'd still like to have a 6 motor drive with one transmission and one gearbox though. So I figured, what if we had the engine outputting to both sets of wheels, but on one set employ the 'reverse gear'. When we're in the normal driving mode where we're swerving and whatnot, the reverse gear won't be engaged. But when we want to steer, we'll pop in the reverse gear causing the wheel sets to spin in opposite directions; and if we want to turn right, we'll run the motors foward, and instead if we want to turn left, we'll run the motors backward.

And we could use software to allow the driver to switch between conventional tank-drive and the swerve drive easily enough through this.

Laff, $@#$@#$@#$@# you Rodd! Looks similar to what I had in mind; except for the steering bit. I kind of want to avoid the whole "car" steering bit... seems like a bit much having to include differentials and whatnot.


It's a good idea to be able to just turn the wheels around 180 degrees to achieve the reverse; but wouldn't you need four motors (one for each wheel) in order to spin two of them backwards and the other two forwards?


And thanks for the wheel idea... I had something like that going around in my head, just couldn't make the parts come together properly. And after seeing that; I got a slightly different approach to it. I'll model it and put it up tonight after work.


EDIT:
Al... does that mean you can't use 6 motors ONLY when turning? Would they work fine for a swerve drive not turning? And then say when we are turning, we could just use software to make two of the motors not run during that time?

[Rod]

Originally Posted by Al Skierkiewicz

I hope that the drive train is not detailed in your drawing. There is a need for slip clutches in the differentials to allow turning. Without them the drive motors go into stall during turns. Just two drills or two Chalupas driving the wheels means 200+ amps during turns. Add the two FP, two Chalups and two drills and you have exceeded the current capabilities of the battery and possibly tripped the 120 amp breaker. There was a GM team who had a similar design without the clutches and they repeatedly set the drill motors on fire during turns using one drill for front and one drill for back. This is one application where current sensing is an eye opener. Sorry.

There are slip clutches in the differentials