How to choose FRC drivetrain wheels: coefficient of friction, Shore A durometer, tread compounds, Colson vs pneumatic vs traction, diameter, and pushing power.
to read
words
sections
Your drivetrain can have the strongest motors in the field, a flawless gear ratio, and a perfectly tuned control loop, and still lose every pushing match and spin its wheels in autonomous. The reason is almost always the same: the robot is asking the floor for more force than the wheels can actually transmit. Wheels are the single interface between all that stored electrical and mechanical energy and the carpet. Get them wrong and nothing upstream matters. Get them right and a modest six-motor drivetrain can shove robots twice its horsepower.
This guide is about choosing that interface deliberately. We will start from the one physics equation that governs traction, work through coefficient of friction, durometer, and tread compounds, compare the wheel families you will actually buy (Colson, traction/plaction, pneumatic, omni, and mecanum), and then get practical about how wheel diameter trades speed for torque, how traction limits protect your motors and breakers, and how tread wears out over a season. By the end you should be able to look at your robot's weight, motor count, and drive style and pick a wheel with reasons, not vibes.
Every traction question in FRC is a special case of the classic friction relationship:
F = μN
Two consequences fall straight out of this equation, and they drive most drivetrain decisions.
First, only the weight actually sitting on your driven wheels counts. If your robot weighs 120 lb but half that weight rests on undriven omni or caster wheels, your traction wheels only get ~60 lb of normal force to work with. This is why serious pushing robots put as much of their mass as possible over the driven wheels, and why weight transfer during acceleration matters — a robot that pops a wheelie unloads its front wheels and loses the grip there.
Second, surface area is not in the equation. A wider or larger-diameter wheel does not, by the idealized model, grip harder — the pressure just spreads over more contact patch. In the real world contact area matters a little because tread deforms into carpet fibers and the model isn't perfect, but the first-order answer is that traction scales with the weight on the wheel and the material's μ, not with how big the wheel looks. That surprises a lot of rookies. If you want more grip, you add weight to the driven wheels or pick a higher-μ tread — you do not just buy a fatter tire.
Because μ multiplies weight, it is the lever you get to choose with your wallet. So it is worth understanding what actually sets it.
FRC fields are carpeted (a tight, low-pile commercial carpet), so almost every published μ figure is "tread on carpet." Values shift on the polycarbonate, aluminum, and steel surfaces that show up as ramps, bumpers, and field elements, and every tread grips smooth surfaces far worse than it grips carpet. Here are representative measured numbers from FRC sources — treat them as ballpark, because μ varies with carpet age, dust, wheel wear, and how hard you press:
| Tread / wheel | μ on carpet (static) | Notes |
|---|---|---|
| Blue nitrile roughtop | ~1.2 | Community-measured; grippiest common traction tread |
| Pneumatic tire, fully inflated | ~1.27 on carpet; ~0.61 on polycarbonate, ~0.49 on steel | Highest carpet grip, but soft and heavy (per Mr. McTavish's measurements) |
| KOP "Hi-Grip" wheel | ~0.95–1.0 on carpet; ~0.31 on HDPE | Kit-of-parts wheel; big drop on smooth surfaces |
| Colson (TPE) | slightly below roughtop on carpet | Beats roughtop on polycarbonate and aluminum |
Two takeaways. Blue nitrile roughtop and pneumatics sit at the top for raw carpet grip, around μ ≈ 1.2. Colson wheels give up a little carpet traction but hold up better when the robot crosses smooth surfaces, because their harder TPE tread doesn't glaze over the way soft rubber can. And notice how far every number falls off smooth surfaces — a wheel that grips carpet at 1.0 might only manage 0.3 on HDPE, which is why robots that have to climb painted or plastic surfaces behave completely differently there.
A practical rule the community repeats: roughly 1.5 inches of Colson tread ≈ 1 inch of roughtop tread in traction, per Chief Delphi discussion. That's a useful mental conversion when you're deciding how wide to go.
Durometer is a measure of how hard a rubber or plastic is. FRC wheels are quoted on the Shore A scale, which runs 0 (marshmallow-soft) to 100 (hard like a shopping-cart wheel). It is easy to conflate durometer with grip, but they are not the same thing — durometer describes the material's stiffness, and grip depends on the compound and the durometer together.
Here is the relationship that actually matters:
AndyMark's compliant wheels are the clearest illustration, because they sell the same wheel in a range of durometers so you can feel the tradeoff directly. Per AndyMark, the lineup runs 35A (green, softest and grippiest), 40A (orange), 50A (blue), and 60A (black, firmest), and the firmer wheels carry a higher RPM rating precisely because they don't balloon and lose shape at speed. For a squishy intake roller you want 35A; for a fast drive roller that must hold diameter, 60A makes more sense.
The famous Colson Performa wheel — a robotics staple — is molded at roughly 65 ±5 Shore A, per the Colson Group specification, with a thermoplastic-elastomer tread co-molded onto a polyolefin core, rated for intermittent service from about −45 °F to +180 °F. That mid-hardness is a big part of why Colsons are so popular: hard enough to last a whole season and hold diameter under a 120 lb robot, soft enough to grip carpet decently and shrug off impacts.
Durometer is not a "bigger is better" number. It is a dial you set based on whether this wheel's job is grip and compliance (go softer) or longevity and speed (go harder).
Two wheels at the same 55A durometer can grip very differently because they're made of different rubber. The compound matters as much as the hardness.
The important habit is to separate the two variables in your head: compound sets the ceiling on grip and wear; durometer sets how the material behaves under load and speed. A soft nitrile and a soft natural rubber are not interchangeable even though both are "soft."
Beyond tread, you're picking a whole wheel architecture. Here's the landscape, and how each interacts with the drivetrain types you might build.
| Wheel type | Grip on carpet | Weight | Best for | Watch out for |
|---|---|---|---|---|
| Traction / plaction (interchangeable tread) | High (with nitrile roughtop) | Low–medium | Tank/WCD drive wheels, pushing power | Tread strips can peel; needs re-treading over a season |
| Colson (solid TPE) | Medium-high, great on smooth floors | Low | Swerve modules, bulletproof tank drives | Slightly less carpet grip than roughtop |
| Pneumatic (air-filled tire) | Highest carpet μ (~1.27) | High | Rough terrain, older-style KOP drives | Heavy, squishy, can go flat, hard to keep round |
| Omni (rollers on the rim) | Low along the roller axis (free-rolling sideways) | Low | H-drive strafe wheels, undriven support, turning aids | Almost no sideways grip — never your only drive wheel for pushing |
| Mecanum (angled rollers, 45°) | Medium forward, allows strafing | Medium | Holonomic drive without swerve complexity | Low pushing power, sensitive to weight distribution and slip |
A few decisions this table drives:
Wheel diameter is a gear ratio you can't see. It sits at the very end of your drivetrain, after the gearbox reduction, and it trades the exact same way gearing does.
Two relationships, both from basic mechanics:
So going from a 4-inch to a 6-inch wheel makes the robot 50% faster and cuts contact-patch force by roughly a third, all else equal. That is identical to lowering your gear reduction — which means diameter and gear ratio are interchangeable knobs, and you tune them together. A common pattern is to pick the wheel diameter that fits the packaging (ground clearance, module size, bumper height) and then set the gearbox reduction to hit your target free speed, rather than choosing diameter for speed directly.
Diameter also changes ground behavior. Larger wheels roll over field seams, cable protectors, and debris more easily and keep more consistent contact on uneven carpet; smaller wheels lower the robot's center of gravity and pack into tighter modules. For most modern robots, drive wheels land in the 3-to-6-inch range: swerve modules cluster around 3–4 inches, tank drives around 4–6 inches. If you're sizing all of this, our gear ratios guide walks through picking reduction and diameter together to land a specific free speed.
Here's where wheels, motors, and your electrical system meet — and where a lot of robots quietly sabotage themselves.
When your robot is stuck against a wall or another robot, one of two things limits it:
You almost always want to be traction-limited, and this is a feature, not a failure. Consider the numbers. A single CIM motor stalls at 2.41 N·m of torque while drawing 131 A, per the FRC motor specs — and a competitive drive runs four to eight motors. If your wheels grip hard enough that the motors never slip, then when you jam against an immovable object every motor drives toward stall current simultaneously. That's hundreds of amps flowing, your 40 A main breakers trip, your battery sags, and in the worst case you cook a motor. (Compare motor limits in our NEO / Kraken / Falcon breakdown.)
Now make the wheels just grippy enough. The robot pushes with everything μN allows, and the instant the load exceeds that, the wheels break loose and spin. Slipping wheels turn faster than a stalled wheel, so the motors stay away from stall, current stays bounded well under the breaker limit, and the drivetrain survives a 15-second shoving match without popping anything. The tread is acting as a mechanical current limiter. This is why maximizing μ and weight-on-drive isn't purely about winning shoves — it's also about controlling how much current your drivetrain can demand. Model it before you build: our current budget tool and the wiring guide help you check that a stall event stays inside your breaker limits.
The design target, then, is not "maximum possible grip." It's enough grip to win the pushing matches you care about, tuned so the traction limit sits below the current that would trip breakers or overheat motors. Add weight over the drive wheels and pick a high-μ tread to push harder; back off if your motors are stalling instead of your wheels slipping.
There's a stability corollary too: the harder you can push, the more likely you are to tip when you suddenly stop or hit something. Grip and tipping are linked — run the numbers with our tipping calculator before you commit to a high-grip, high-CG configuration.
Tread is a consumable. Over a competition season a set of drive wheels can lose a meaningful fraction of their grip, and the failure is gradual enough that teams often don't notice until the robot starts losing shoves it used to win.
What to watch for and do:
Budget for spare tread the same way you budget for spare batteries. Blue nitrile roughtop is sold by the foot precisely because teams re-tread regularly, and having a fresh strip ready in the pit is the difference between a five-minute fix and a lost match.
Put it together with a short decision process:
Do this and your wheels stop being an afterthought and become a tuned part of the system — the part that decides whether all your motors, gears, and code actually reach the floor.
Neither is universally better; they optimize for different things. Blue nitrile roughtop has slightly higher carpet grip (μ around 1.2) and its tread is replaceable, which suits pushing-focused tank drives. Colson wheels grip a touch less on carpet but hold their diameter precisely, last for years, weigh little, and perform noticeably better on smooth surfaces like polycarbonate and aluminum — which is why they dominate swerve modules. Roughly 1.5" of Colson equals 1" of roughtop in traction. Choose roughtop for maximum pushing power, Colson for durability, weight, and consistent odometry.
No, not by itself. Traction is F = μN — coefficient of friction times the weight on the wheel — and wheel size isn't in that equation. A larger-diameter wheel spreads the same weight over a bigger contact patch without increasing total grip. What a bigger wheel does change is the speed/torque trade: it makes the robot faster but reduces pushing force at the floor, exactly like lowering your gear ratio. To actually get more traction, put more weight over the driven wheels or switch to a higher-μ tread.
For drive wheels you generally want a firmer material that holds its shape under a heavy robot and wears slowly — think Colson's ~65 Shore A, or 60A-and-up compliant/traction wheels. Softer wheels (30–40A) grip and cushion better but squish under load, roll with more resistance, wear fast, and top out at lower RPM, which makes them better suited to intake and manipulator rollers than to drive wheels. Match the durometer to the job: hard for drive, soft for grabbing game pieces.
Your drivetrain is torque-limited instead of traction-limited: the wheels grip so hard they never slip, so when you shove an immovable object the motors drive toward stall — a single CIM stalls at about 131 A — and the combined current trips breakers or cooks motors. The fix is to let the wheels slip before the motors stall. Being traction-limited caps pushing force at μN and keeps the wheels spinning, which holds current below the stall value. Tune grip and drive weight so a full-stall shove slips the tread rather than stalling the motor.
Treat tread as a consumable and inspect it every event. Replace or re-scuff when the roughtop texture has flattened, the surface has glazed shiny, or the wheel diameter has visibly shrunk (which also throws off encoder odometry). Teams running plaction/traction wheels re-tread with fresh blue nitrile strips regularly and keep spare tread in the pit — it's a five-minute swap. Rotating wheels between positions evens out wear, and simply wiping dust and debris off the tread between matches restores grip for free.
This article is AI-assisted: drafted from primary sources, then reviewed and edited by hand. Errors still get through. When one is reported we fix it and write down what changed — publicly, in the corrections log.
Keep going
You came here for one answer. These lessons teach the same subject properly, in the order a team actually learns it. All 47 are free and none of them need an account to read.
Go deeper
LearnFRC has 394 free FRC lessons across every department. Create a free account to save your place, track your progress, and earn a certificate.
Create a free accountKeep reading
Structured lessons and quizzes across every department. Create a free account to save your progress, track your team, and earn a certificate.