Treaded tank drive vs. Traction wheels

We are currently having a debate on our team over traction wheels vs. treaded tank drive (a.k.a. track drive) and there are three arguments:

Treaded tank drive has more traction (Given same weight, motors, wheelbase, and coefficients of friction)

Traction wheels are more prone to slipping (given same weight, motors, wheelbase, and coefficients of friction)

If you get rammed by another bot, 4 traction wheels would “bounce” and have less traction than treaded tank drive (given same weight, motors, wheelbase, and coefficients of friction)

I am arguing that none of these are true, and that if you have the same robot with traction wheels and treads (given same weight, wheelbase, and coefficients of friction) they would have identical traction and the pushing match would come down to who has more motor power. Frictional force does not vary with surface area, so I see no reason that you would have weaker traction with 4 traction wheels.

Do treads have significant advantages (excluding obstacle clearance) in matches when it comes to defense/ramming/pushing? I would think 4 traction wheels (with identical coefficients of friction to said treads) would be equally good for defense/ramming/pushing but I haven’t run either so I don’t know if there’s something I’m missing.

Either are perfectly fine to use, but traction wheels allow for easier maintenance and adjusment. Also, I believe treads have the possibility of breaking, rendering the drivetrain useless on the field. I don’t think that the traction is the biggest factor when deciding what to use. Overall I think that traction wheels are the better choice for FRC applications.

Traction isn’t entirely independent of contact surface area when the tread interlocks with the carpet (e.g. roughtop, really any local deformation, etc). That said, the efficiency and reliability losses in going with a tank tread setup far, far, far outweigh any potential gains in traction they offer. It’s just not worth it, when more / wider wheels get you almost the same benefit if you really truly need every ounce of traction you can muster.

You should take a good look at the 2016 FRC Game. Most all of the best robots in the world were able to use pneumatic wheels in a tank drive formation to effectively navigate throughout the “roughest” terrain that FRC has ever had.

As for the pushing match scenario, any performance differences between tank tread and tank drive are marginal at best. Factors such as driver skill/positioning of the robots in a “pushing match” are much much more important.

Tank drive with wheels may actually have better resistance to lateral pushing because the wheels will “sink” into the carpet a bit as a result of putting the same amount of weight on a smaller surface area of the carpet.

If the coefficient of friction is the same then there is no advantage in a pushing match.(Providing weight, power, etc. are the same)

Even if the tread interlocks, wouldn’t two robots with the same material on their traction wheels/treads have the same coefficients of friction with the carpet? I could see some local deformation affecting friction a little bit, but I doubt it would be by a measurable margin, let alone enough to change the outcome of a pushing match.

Interesting. I didn’t think about this. Do you think the smaller surface area makes them more prone to skipping/scrubbing/bouncing as a result of a ram?

A wheeled tank drive will be more reliable then treaded drive as it’s mechanically simpler. If your drive train breaks your finished for that match.

With dropped centers, a 4 wheeled drive will turn at least twice faster than a tank drive.

Stronghold featured a huge number of tank drives, but large pneumatic tires robots worked more effectively, and typically were faster.

Yes and No. It get’s weird with wheels because in some cases, their surface area is so small it actually changes the coefficient of friction. (They dig in to the surface giving them an advantage.)

Well, let’s take these in order, then.

These two are actually the same argument, but the statement indicates some confusion: if the drives (identical in all other meaningful aspects) are assumed to have the same coefficient coefficient of friction, then they must have the same traction. This is definitional - it’s what the coefficient of friction “means.”

Now, a better statement might be that tank treads are likely to have more traction than traction wheels made out of the same material - and this is likely to be true, though the size of the effect is complicated and hard-to-predict and if you’re really interested in super-high traction you’re probably best-off using something like W-treaded VersaWheels and being sure to swap them out often.

I will note, though, that traction alone is usually pretty low-down on the grand list of “crucial drive concerns,” and traction alone is a very poor reason to choose treads over wheels.

If you get rammed by another bot, 4 traction wheels would “bounce” and have less traction than treaded tank drive (given same weight, motors, wheelbase, and coefficients of friction)

I have no idea where this idea came from (what specifically is meant by “bouncing?”), as I’ve never heard or seen anything that would lead me to make a claim like this.

I won’t say that this is false, but I will say that it’s something that you definitely shouldn’t believe without compelling evidence (personal intuition does not count as “compelling evidence,” for the record).

I am arguing that none of these are true, and that if you have the same robot with traction wheels and treads (given same weight, wheelbase, and coefficients of friction) they would have identical traction and the pushing match would come down to who has more motor power. Frictional force does not vary with surface area, so I see no reason that you would have weaker traction with 4 traction wheels.

The idea that “frictional force does not vary with surface area” is a first-order approximation that breaks down in surprisingly common situations, with FRC wheels on carpet most definitely being one of them.

However, as I mentioned before, the size of the effect is hard to predict and if “we need to eek out a bit more effective traction” is a major drive requirement, a) you’re probably approaching drive design with the wrong priorities (unless you have a really good and really specific reason for wanting that extra pushing power), and b) picking tank treads is likely far from the best way to satisfy that requirement anyway.

Do treads have significant advantages (excluding obstacle clearance) in matches when it comes to defense/ramming/pushing? I would think 4 traction wheels (with identical coefficients of friction to said treads) would be equally good for defense/ramming/pushing but I haven’t run either so I don’t know if there’s something I’m missing.

The additional traction from treads (compared to wheels made out of the same material) may provide a slight benefit in certain situations during very physical play, but it will be absolutely dwarfed by the effects of other, far more important drive design concerns (robustness, proper gearing, efficiency, etc).

Pushing matches between two traction-limited tank drives have an overwhelming tendency to end in a stand-still, or else have their outcome determined primarily by a combination of driver skill and happenstance. A small increase in your available traction is likely not going to make worlds of difference in your success in pushing matches, and successful play rarely depends on winning an outright pushing match, regardless (as a defense bot, pushing another robot too far can take you out of position, while as a bot trying to overcome a defender, it is almost always faster to find some way to go around), and if you really do decide you want that extra oomph, wheel selection is probably a way better way to achieve it than treads…

At BC18, team 885 had treaded tank, and 3 matches in elims, their treads slipped and were rendered useless.

Just my 3 cents

That’s true in theory, as any students who’s taken a basic physics class can tell you. But like Chris said above, the reality is that contact area does matter when you’re dealing with interlocking surfaces (like tread on carpet). It’s no coincidence that a robot with wider wheels will win a pushing match against a robot with thinner wheels cet. par. That being said, the traction advantage you gain is minimal compared to all of the other factors involved. Driver skill (which tends to be negatively correlated to robot complexity) is a much larger factor IMO. And also, contact area for treads often isn’t the entire length of the tread because the entire tread isn’t covered by idlers, so some tank drives will have more contact area than treaded drives.

The reality is that treads are good in theory, but often they are less-than-equal in practice. There are a lot of failure points that need to be accounted for in order for it not to fail, let alone be more useful than a regular tank drive setup, and there is not a ton of documentation on best practices or an easy COTS way to build one. In that sense, I would almost compare it in complexity to swerve, with much less benefit.

Take turning for example. A decently designed 6+ wheel tank drive has no problem turning. A treaded design, however, has the possibility of throwing treads and draws more current, therefore it is more likely to pop breakers and cause brownouts. I’m not saying that it is impossible to build a good treaded drive, nor am I saying that there are no benefits to one, but I would argue that a solid, easy-to-build tank drive is a much better option for a majority of teams and games. No matter what, I would definitely recommend prototyping it in the offseason before you try it for real.

EDIT: If you want the benefits of a tank drive without so many of the negatives, I’d look at using a lot of small wheels close together, like 271 or 341

It’s important to remember that the friction models we learn in high school (and in college for most people) are based on two rigid objects interacting. Once you introduce object compression and varying degrees of density (like with carpet and tread for wheels/tracks) the models get more complicated.

Also, you say one may be more prone to slipping than the other… that is entirely dependent on the setup and gear ratios involved. You can make tread be traction limited, and you can make wheels stall out, or vise-versa. There’s so much more that goes into a design than answering the question of “wheels or treads?”

You’ll probably get a lot of fairly accurate impressions on here, but no real, solid mathematical proof that one is better than the other. The best way to tell is to get a chassis and equip it with treads and traction wheels. Set up a series of tests you can perform with each configuration (pushing power, turning speed, top speed, acceleration, etc) and collect results. Look at things like power output (use a force gauge if you can) and power input (record the current going into the motors, along with the known current battery voltage). Let us know what you find, and see how those results compare to any theoretical models for friction you have!

Good question. If you build wheeled tank drives with a few key ideas in mind, scrubbing and bouncing shouldn’t be an issue.

In your initial post you describe a wheeled tank drive as having 4 wheels. Most teams use six/eight wheels and “drop” the center wheel(s) lower than the outer counterparts (Not by much, maybe only an eighth of an inch for a 32" long robot.) This way the robot “rocks” back and forth (only a tiny bit, if it’s done by a small amount you barely notice.)

The type of wheel plays a big part in this as well. The larger pneumatic wheels https://www.andymark.com/8-inch-pneumatic-wheel-p/am-0970.htm definitely have more scrub than the HiGrip wheels often found in the KOP. http://www.andymark.com/HiGrip-Wheels-p/am-higripwheel.htm?1=1&CartID=0

If you want to really learn about drivetrain design principles, check out the seminars/videos I’ve linked below:

https://www.youtube.com/watch?v=ADcqUWYyHgk

https://www.youtube.com/watch?v=6lmKLTZZLBE&list=PLIY-TB1MAu-X9ZcNqt-ot6_JM2Z02zZ6L&index=16

Good drivetrains are deceptively hard to make, but if you can get something reliable together and practice with it a lot, you’ve already got a big advantage over most new teams.

Feel free to message me if you have any other questions.

We’re contemplating octocanum, so we’d have 4 wheels :slight_smile:

Thanks for linking all of these. Will be sure to show team

I really want to test both… but we are looking to purchase one drivetrain to test out/build on over the summer (for possible use next year). We also are going to be buying Talons, sensors, etc. to play with over the summer so the budget at the moment doesn’t give us enough room to invest in multiple drivetrains for testing

In a normal situation, the contact patch of the tread does not increase traction because as the contact patch increases, the pressure at each particular area of interest decreases.

Where this starts to fall apart is when the materials interlock, because the materials “grip” onto each other until either the material fails entirely or until one material deforms enough to “release” its grip on the other.

Consider a really exaggerated example - soccer cleats in grass. Compare a shoe with one cleat to a shoe with 20 cleats. Which is going to grip better? The shoe with 20 cleats does. This is because all of the load of the shoe being concentrated in the 1 cleat, leads the turf to fail more quickly, meaning it loses its grip with less applied force. This load spread out on 20 cleats limits the local deformation / failure at any particular area.

On a smaller, less extreme scale, this same principle applies to roughtop tread and carpet. The “hooks” in roughtop tread can, with a small enough wheel, “dig in” to the carpet, enhancing traction. In these cases, it’s been observed that in 4" wheels, widening the wheel will significantly improve traction. Even for wheels that don’t interlock with the carpet, if there is the possibility of some localized tread failure (imagine a wheel made up of pencil eraser rubber), the wider wheels will fail less quickly and thus maintain traction better.

For the most part, these effects are very marginal - the only way to know if you have a specific edge case where surface area makes a significant difference is to do testing.

Definitely skip the tank treads then. There’s a reason they are so rarely used - and a team like yours is better off getting the basic, fundamentally solid competitive tank drivetrain down than to spend time with something relatively niche and gimmicky.

I think a better term would be skipping…
Say you are getting pushed from the side. The wheels would “skip” once the force static friction is overcome.

TIL. I had just assumed good ol’ F = u*N was valid for nearly all scenarios.

What do you think about wheels with tread riveted to them?

Thanks for the response

You don’t really need 2 whole drivetrains, just 2 sets of wheels. Get some 4" treaded wheels and some 4" tank tread pulleys. Switch between them for your testing. Of course it’s more difficult and a bit more money than just building one, but it’s good experience and good testing.

The set-ups for a good tank-tread drive and a good wheeled tank drive are sufficiently different that this is likely a bad idea.