08-15-2008 02:25 AM

Joe G.

Re: pic: Omnibot

Interesting design. Several questions

-What, if anything, will be used for suspenson?
-The gearboxes look custom. Details? Ratio? Material? etc.
-How much does it weigh?
-Welded or bolted frame?

08-15-2008 06:15 AM

Schnabel

Re: pic: Omnibot

HAHA, not to make fun of your design as it looks really good, but the first thing I thought of was that this would make a great demo-bot because you could make it spin in circles very easily!
Now on a more serious note, it looks like you made it so the front and back wheels could move to be supported by the outer frame. Is this true?

08-15-2008 09:38 AM

Daniel_LaFleur

Re: pic: Omnibot

I believe you will see tipping in the corners of the robot, especially when changing from forward motion to strafing or skittering. You may want to move the motors and omni wheels to the corners of the inside structure to alleviate that.

Other than that, looks very good and manuverable.

08-15-2008 10:03 AM

Andy Baker

Re: pic: Omnibot

Great job on this design.

I think that you are intending to keep the 4 wheels in a square pattern, with the distance between the faces of each pair of wheels being equal. You don't have to do that... you can take the wheel on the left and the wheel on the right and move their mounts out toward each side.

Sincerely,
Andy B.

08-15-2008 11:46 AM

sgreco

Re: pic: Omnibot

If you want, another thing you could do is put small, lightweight casters on the corners to prevent tipping.

08-15-2008 04:15 PM

kajeevan

Re: pic: Omnibot

ok specs on this design are:

being an omnibot i believe speed is very much important so it goes 16f/s.
wieght is about 40lbs but i can bring it down to about 35-36 i believe.
the gearbox is custom but is really simple im using many toughbox gears and axles but the ratio is changing in order to get up to speed.
yes im trying to keep it square but if you say moving the wheels out to the edge wont change the performance im willing to do that.
aslo yes everything on the base is bolted im following my mentors advice when i was in grade 9 saying that if it can be bolted bolt it because if a weld brakes during competition theres no way to fix it

if we don't need total length i do plan on keeping the base 26'' by 26''.

08-15-2008 07:31 PM

Tristan Lall

Re: pic: Omnibot

Just a few things:

• With the 4-wheeled omni-drive you've got there, you're capable of getting at most 4 × P_CIM ÷ √2 watts of output power in a straight line, because of the geometry (when 4 wheels are at 45° to the instantaneous direction of travel). But you'll still consume the full amount (4 × P_CIM). You can only get full output power when spinning (which is probably useless). That's a big enough performance penalty that you should weigh the increased maneouverability against the decreased efficiency. (It might not be the best for a game with a lot of pushing of things, for example.)
• Ground clearance and stability are of concern when the sides overhang like that. Note also that if you use casters to support the corners, whenever they're exerting force on the ground, that normal force is no longer available at the drive wheels, and consequently you have less traction. That might be important if you need to push something.
• If you do go with an asymmetrical design (like Andy and others had mentioned), the math is a tiny bit harder because the wheels are no longer equidistant from the geometric centre. This is OK, because for fancy motion (other than straight lines and spins), your centre of rotation is going to move around (i.e. it won't necessarily coincide with the geometric centre). Seriously consider a way of implementing that fancy motion, because it allows an additional degree of freedom to be used at any given time. (And the controls guys need to be thinking about an interface for it ASAP....)
• That's going to weigh too much...the 40 lb estimate seems optimistic. Consider a design where the bumpers form part of the structure (though it means the bumpers have to be precise, strong and rigidly attached, it also means that you're not duplicating structure). Also look at supporting the wheel axles on plates that mount to the back of the bumpers—I'd say you can eliminate six lengths of box beam that way. (Maybe the support plates can be birch plywood, too, for weight reasons.)
• Can the box beam be replaced with C-channel? It's generally easier to install and remove things from a piece of channel, because there's no groping around inside. (Especially important when using bolts and nuts.) And there's no need to machine out clearance holes. The tall direction of the beam is the strongest, anyway. As a result, there's likely little need for the second vertical web.
• When using bolts at the corners, you often need to put more than one at each joint. Otherwise, it acts like a pinned connection, and doesn't resist angular motion (so the frame might parallelogram*).
• Consider the tradeoff between the weight of the Toughbox parts, and their margin of safety versus the loads expected on this gearbox. (They're too strong!) You might consider smaller gears, especially for the first reduction. Of course the Toughbox gears are very easy to assemble, given a shaft with a hex section. Either way, plan to lighten the gears, as a matter of good practice.
• Depending on the flatness of the event floor, and the design of the field, you may run into issues where three wheels touch the ground, but the fourth doesn't. That will make control troublesome with a 4-wheeled omnibot, or a Mecanum system. (At GTR and Waterloo, the floors are very flat, even with the masonite. At the Championship, this is absolutely not the case, because of the plastic tiles laid down underneath.)

If you want to discuss any more of this, I'll be in Toronto over the weekend, and maybe early next week.

*Karthik, I checked: this is sometimes used as a verb in engineering literature. I'm not making things up.

08-15-2008 08:50 PM

Richard Wallace

Re: pic: Omnibot

Engineering literature?

Is that like jumbo shrimp, military intelligence, or legal brief? You know, an oxymoron?

Seriously, those (Lall: supra) are good suggestions.

08-17-2008 06:18 PM

joeweber

Re: pic: Omnibot

We did an omni bot very similar in 2004 and won Xerox creativity award. You can view pictures at http://www.team1322.org/robotics_03-04.htm . What we did is put the wheels in the corners so you can push them way out form center. We also used three omni wheels at each drive for more traction. Back then we had to build our own omni wheels http://www.team1322.org/omni_drive.htm . The front of our robot was your corner. The robot was very fast and turned on a dime. I do believe the Mecanum drives would be more effective for you. It is more stable going straight and dose all that the omni bot does. Nice drawing and a grate design that is fun to drive. We still get ours out and drive it all over.

08-18-2008 11:58 AM

Tazlikesrobots

Re: pic: Omnibot

We used a design very similar to yours this year and it worked very well for us! Our chassis was square shaped and the wheels were placed as close to the edge as possible.

08-18-2008 12:22 PM

XXShadowXX

Re: pic: Omnibot

You will most likely be pushed around alot with your current design (this is from my personal experiance). But omni's mounted to and prependicular to the direction of movement could spell trouble.
Also you don't need to mount your wheel in a perfect square, if you mount what I'm guessing is your forward and rear omni futher back you will increase stabbility. Just make sure your wheel that are found on the same axis (left and right, forward, and, rear) are mounted so that they lie on the same circle. If your really good you should be able to make both of your circles have the same center point (and center point of robot), but good luck...

08-22-2008 07:46 PM

gimpnub

Re: pic: Omnibot

why don't you use a three-wheel setup instead of a four-wheel setup?

correct me if i'm wrong but i think it would give you more stability and reduce your weight at the same time

your only difficulty might be programming it with 0, 60, and 120 degrees instead of 0, 90, 180, and 270 degrees, but there are also less wheels to program

i don't know how speed would be affected by the 3-wheel setup

if i'm wrong in any areas please forgive me and ignore my comments

08-22-2008 08:12 PM

Joe G.

Re: pic: Omnibot

The main advantage of four-wheel omni that I can think of is power. Even with vectors doing crazy things to the power output, 4 motors is stronger than 3.

08-22-2008 10:52 PM

Protronie

Re: pic: Omnibot

Very interesting... is this something that going to be built or just a design exercise ?

-p

08-22-2008 11:11 PM

EricH

Re: pic: Omnibot

Reduce weight, yes, but 3-wheel setups have to be carefully thought out. I can only think of 4 off the top of my head: 330 (2002) had a triangular robot and no omnis; 67 (2005) had a peculiar triangular setup that I don't quite remember, but it involved flopping down; 16 (2006) had a 3-wheel swerve with pads to keep the frame from hitting the carpet and digging in, and 148 (2008) had a 3-wheel swerve patterned off of 118's V6 and a much smaller robot than the other three. The only one of those that involved omnis was 67's, IIRC. I remember hearing that 67's drive code took up an awful lot of room on their controller...

More stability is debatable. If you've got it fully in a square or circle, possibly. But with a frame like this, I don't think so. You've got a big risk that one corner will go down and dig into the carpet.

Speed might not be affected. I'd have to do the vectors to figure that out, and I'm not in a position to do that right now...

08-23-2008 12:18 AM

Aren_Hill

Re: pic: Omnibot

Im almost completely positive 67 in 2005 was a 3 wheel swerve with no omni's involved.

you can see the front two wheels here : http://www.chiefdelphi.com/media/photos/20772

08-23-2008 12:43 AM

Joe G.

Re: pic: Omnibot

857 did a 3 wheel omni, or "kiwi drive," in 2002

http://www.chiefdelphi.com/media/photos/13495

08-23-2008 01:49 AM

kajeevan

Re: pic: Omnibot

Right now its just a design but who knows I may just get to build one.

08-25-2008 01:22 AM

EricH

Re: pic: Omnibot

I'm pretty sure it was a kiwi-type system, like below. I tried to search, but no luck yet.

08-25-2008 01:27 AM

AdamHeard

Re: pic: Omnibot

Nope, sorry Eric, but 67 definitely had a 3 wheeled crab in 2005.

08-25-2008 01:31 AM

EricH

Re: pic: Omnibot

Actually, I found a picture... of 67 2005... in a match... on two wheels, showing the third. NOT a true swerve. Swerve, yes. Three wheeled robot, yes. But not a three-wheel swerve. One wheel steered, three drove. Zoom in on the picture. One wheel is turned; the others appear not to be able to.

08-25-2008 01:39 AM

AdamHeard

Re: pic: Omnibot

Actually, I found a picture... of 67 in 2005...during build... on the ground, showing all three wheels. A true swerve, with all three wheels able to rotate.

No hard feelings eric

08-25-2008 02:16 PM

seanwitte

Re: pic: Omnibot

If you move the wheels to opposite corners the math will work out exactly the same as it would for the configuration shown.

There are two advantages that a holonomic base built using this configuration has over a swerve drive:

1) The modules are lighter and less complex.
2) The robot can translate and rotate at the same time (I've seen this called "frisbee motion")

The programming is more complex, but there are plenty of examples available to get you started.

08-25-2008 11:22 PM

EricH

Re: pic: Omnibot

And rotate independently... No wonder the code took so much space in the controller.

Back on topic: If you're going to do 3WD, plan it out first. Carefully. As noted, only one robot (to our knowledge) has had a 3WD omni system; most of the other 3WDs were crab/swerve systems.

08-25-2008 11:30 PM

Andrew Schreiber

Re: pic: Omnibot

Also, the one robot that did a 3 wheeled omni (857) didn't use code to control their wheels, they used hardware (http://www.chiefdelphi.com/media/photos/16410) Dont forget that sometimes that solution works better than complex code. No reason to make the code more complex than it needs to be. Could it be done in code? Yes, is complex code always better? That is up to you.

08-26-2008 08:15 PM

Dowjonesbotics

Re: pic: Omnibot

Cooool, are those the robots that spin really really fast??

08-27-2008 05:05 PM

1jbinder

Re: pic: Omnibot

Hi
I like the design. Both my team(852) and an alliance partner at Davis 2144 built bots like this. They work well but can be a challenge to program. autonomous is almost impossible without a well coded gyro. Anyways give it a try and see what happens. That is what we did.
Julian

11-09-2008 12:05 PM

daltore

Re: pic: Omnibot

Kiwi drive/Killough platforms are generally not any more stable than the 4-wheel holonomic drives. The arc lengths between wheels are larger than for the 4-wheels, meaning more area to tip. As for spacing the wheels out of a perfect square, yes, it works, but buy moving them out one more inch, you will require code of much greater complexity because of the angle calculations you will have to use to get the same effect. With the new control system, that shouldn't be a problem memory and speed wise, but it's just more things to crash in the program. Nice design, it's pretty!