Casters vs. Unpowered Omni wheels

[nnfuller]

So I have been looking at a lot of older robots for inspiration for a new chassis design before build season and I was wondering if any veterans could tell me why we don't see casters anymore. I have seen a large number of robots with unpowered omnis, but no casters... Why do you think this is, becase from where I'm standing...

Casters
Pros:

Self Contained
Pre Assembled
Bolt and Go Simple
Smooth Ride

Cons:

Have a "Direction" that must be overcome
Cannot be Powered

[Andrew Schreiber]

So I have been looking at a lot of older robots for inspiration for a new chassis design before build season and I was wondering if any veterans could tell me why we don't see casters anymore. I have seen a large number of robots with unpowered omnis, but no casters... Why do you think this is, becase from where I'm standing...

Casters
Pros:

Self Contained
Pre Assembled
Bolt and Go Simple
Smooth Ride

Cons:

Have a "Direction" that must be overcome
Cannot be Powered


Because casters tend to change the direction randomly depending on how they start. Ask any FLL team.

[ratdude747]

casters tip easily. no matter what direction they are pushed, they always pu the pivot point (the axle) in a bad position in terms of stability.

not that casters never work, but that i have only seen one robot ever that used them well , and that was in a picture (some veterans around here would be able to guess it). however, I have seen many examples where casters were a complete failure.

[Koko Ed]

We used casters on our 2008 robot for the front wheels. They worked pretty well.

[davepowers]

GUS' very first year we used casters, they worked for us, sort of. I do remember the robot did a very good job of not going straight, that might not have been the casters though. There were a few problems with turning, sometimes the casters would turn a way they shouldn't, but it would always push through.
http://team228.org/gallery/1/slideshow/MVC-004S.jpg

-Dave

[IndySam]

We used casters early on until 2005. We now have a hard and fast rule, no casters allowed on the robot, ever.

[Billfred]

1293 used casters its first two years. The first robot (whose autonomous was just timing and held breath) would be thrown off-course if the casters weren't right but was otherwise underpowered enough that it wasn't a huge issue. The second one (powered by CIMs) verged on uncontrollable on the field, easily the weakest link in its design. They haven't looked back, going with treads ('07, '10), N-wheel-drive ('06, '09, '11), or mecanum ('08).

2815's always driven all the wheels on the floor, so no experience there. (I would want to know why someone wouldn't power an omniwheel; there are certainly some scenarios, but I can't see those scenarios being anything but a teensy minority.)

[SenorZ]

We used four casters on the corners, and two powered wheels centered on the sides. Using a tank drive program it worked quite well.

Pros:
1. Cheap!
2. They work!

Cons:
1. Wobble while they orient themselves
2. Look lame

[EricH]

It depends somewhat on your application and your specific type of caster. For example, 330 uses a lot of pneumatic caster wheels (modified to be driving wheels, that is!)

In general, omni wheel > ball caster > other caster types. But, that does have a few caveats.

First, you hit a major con of casters: You cannot power them. Well, not easily. With an unpowered omni wheel, powering is as simple as adding a sprocket and chain to the rest of your drivetrain, and you gain some forwards traction. Every time you add an unpowered wheel to your drivetrain, you lose normal force on the powered wheels, which loses you frictional forces, which just so happen to provide traction. Long story short, loss of traction. By placing casters, you can lose traction.

A fairly large advantage of omnis/con of casters is that omnis can almost literally drop into a drivetrain with one bolt (the axle) and no frame modifications besides a hole for said axle. (OK, so bearings in the omnis... but close enough.) A caster needs to have multiple mounting points matching the base pattern, which is a little bit more involved and takes some minor planning.

And now we get to the other con listed by the OP. For a pre-assembled, cheap (one he forgot in the pros), smooth-riding, easy-to-mount (well... see previous paragraph) solution, you get: Unpredictable driving motion. You can insert your own scenario where you need predictable driving motion, I'm sure! But with omnis, you know what's going to happen for sure, whenever you apply power in a given manner.

As a point of comparison:

Omni Wheel Pros:
-Can add power
-Predictable motion
-Can be used as a sensor wheel when unpowered (casters can't do this as well)
-Drop-in

Omni Wheel Cons:
-Cost
-Fewer sources
-Bumpy ride (depending on type)

Caster Pros:
-Self Contained
-Pre Assembled
-Cheap
-Smooth ride
-Easy to mount with planning

Caster Cons:
-Unpredictable motion
-Extremely difficult to power
-Tougher to mount without planning
-Harder to use as a distance sensor due to said unpredictable motion

FWIW, I've seen it said on CD that a peg leg is preferable to a caster wheel.

[Ninja_Bait]

Casters probably make sense for a drive like this (http://www.chiefdelphi.com/forums/showthread.php?t=96138).

Not that we have experience with casters - we've used omnis when we had to, and we powered them like 2815. Why do anything else?

[Marc S.]

This (http://www.chiefdelphi.com/media/photos/12952) was a Fully Omni-Directional robot that successfully worked during competition.

It was very successful and arguable the best robot in 2000. Still we have to remember that the competition has since changed, and drive capabilities have greatly improved.

[Chris is me]

This (http://www.chiefdelphi.com/media/photos/12952) was the first FIRST Fully Omni-Directional robot to successfully work during competition.


Slight correction: 47 had an omni-directional robot in 1998 that successfully worked quite well. They won three regionals that year - I don't think there were more than a dozen.

[Marc S.]

Slight correction: 47 had an omni-directional robot in 1998 that successfully worked quite well. They won three regionals that year - I don't think there were more than a dozen.

Here's (http://www.youtube.com/watch?v=_FJFbvHRyco) one of there matches. Oh though your right, upon further inspection they did build one in 1998(back when there were only 5 regionals).

[Al Skierkiewicz]

Casters don't work for autonomous since they have a mind of their own. They are simply a bad idea for any game that has ramps or some other drive over hindrance. They generally do not turn within their own radius so the robot will veer of course. And...have you never driven a cart at a store where the fronts wobble?

[Ninja_Bait]

Casters don't work for autonomous since they have a mind of their own.

That wouldn't be a problem if they were also intelligent. ;)

[Marc S.]

If it was a line following bot you could get away with autonomous. But no guarantees.

[Jon Stratis]

For what it's worth, back in college, I took an Autonomous Robotics class - essentially, we played with MIT's HandyBoard and a bunch of Lego's all semester, with a game/competition at the end of the semester. I ended up designing a caster wheel (out of the provided Lego's) to use on the robot (2 drive wheels in front, 1 centered caster wheel in back) The biggest problem we had with the wheel, before we got it designed right, was it pulling apart or pulling out of the joint where it attached to the robot - there was simply too much force on it when the wheel was trying to turn (and we designed for 90% of the weight to be over the drive wheels)!

Now, given the situation, the caster wheel provided us with a great advantage - Unlike all the other robots (which had 4 traditional drive wheels - no one tried doing a 6-wheel drop center... hadn't even heard of it at that point), ours didn't have to overcome scrub on the wheels when it was turning, and we could turn around the center point between our drive wheels. This simplified other aspects of the design, as the robot could rapidly spin in place to identify the polarized guide lights, while other teams had rotating sensors and then had to line the robot up with the sensor.

Based on that experience, I say casters do have their purpose... but it's going to be a very rare game/robot design where it provides a benefit to a FIRST robot, given all the other options we have (dop center, omnis, etc).

[apalrd]

We once built a 2wd occra robot with rear straight casters and front skids. It was very back heavy, so the skids rarely hit the ground.

We quickly replaced the rear wheels with lunacy wheels, driven.

Don't use casters. Use driven omni or lunacy wheels.

[Superllama12]

Because casters tend to change the direction randomly depending on how they start. Ask any FLL team.

Ha ha, I remember we had so much trouble with that, until we switched to treads, but then we had even more problems with that! But I would definitely rather use unpowered omnis if we had the budget, cuz they don't change directions as easily from bumps. Also, casters cannot deal with uneven terrain, which is a little easier on omnis. My favorite type is mech, because it actually tank drives when you need it, and is much simpler.

[P. Fuller]

Casters are a pain in the rear to work with (in my opinion). They swivel around, which can mess up autonomous for the robot. Also, if there are bumps in the field again (like 2010), casters are much more likely to break than a slick wheel from 2009 (which has the same effect as an unpowered omni). Overall, slick wheels are an easier and much more reliable solution.

[EricH]

Also, if there are bumps in the field again (like 2010), casters are much more likely to break than a slick wheel from 2009 (which has the same effect as an unpowered omni). Overall, slick wheels are an easier and much more reliable solution.
The 2009 slick wheels, though, have a much higher CoF in the sideways direction than an omni wheel (which is near zero, due to the sideways rollers). So, they will drag, which makes turning that much harder. While it won't be as bad as a 4WD all-traction setup (which could be really bouncy, depending on traction), it still won't quite perform as well as an omni wheel, powered or not.

[pfreivald]

We used casters early on until 2005. We now have a hard and fast rule, no casters allowed on the robot, ever.

We used them in Rack and Roll. It remains to this day the most disastrous engineering decision we've ever made. (And we've made some doozies!)

[tlivingd]

We used them in 1997 and I was a driver. They work fairly well if the rest of the machine was designed to work that way and the use a more sweeping style of driving than sudden direction changes. We even had mixed control with single joystick instead of dual joystick tank drive.
Our machine was a front wheel drive tail dragger. It had a forward heavy CG and would scoop tubes and reach forward.
The event was Toroid Terror and the event was in a hexagon with a center mounted goal.
The year prior we used a 2wd center mounted with solid pads instead of casters in the 4 corners. it was tank driven and was fine, just rocked a little.


However in this years event there is no way they would work with the ramps/tables.