Wedgetop vs. Roughtop

[Tristan Lall]

Originally Posted by zaphodp.jensen

The difference in traction from roughtop and wedgetop is going to be stronger. Roughtop has individual "fingers," which increase the surface area in contact with the rough carpet, while the wedgetop only has the wedge pattern, which is going to have to bend as one. This will decrease the amount of plaction in contact with the ground.

What the tires look like (Surface Contact View)
Roughtop:
- - - - - - - - - - -
- - - - - - - - - - -
- - - - - - - - - - - -

Wedgetop:
X
X
X

Or: Think of it this way: Which is going to have more surface area when pressed against the ground with ~120 pounds on it: A Pine tree branch, or a dowel rod?

The pine branch's needles are going to compress and increase the surface area in contact with the surface, while the dowel rod is going to not compress as much, and have less area in contact with the ground.

Good thinking, but it's a little more complex than that.

First of all, roughtop and wedgetop treads come in different materials. Polyurethane [Edit: The Brecoflex Supergrip I was thinking of is PVC tread on polyurethane backing], synthetic rubber, natural rubber, etc.—so it wouldn't necessarily be reasonable to compare a soft wedgetop with a hard roughtop.

If you're just modelling friction between smooth surfaces, friction is independent of surface area in contact (purely a function of normal force and coefficient of friction). But FIRST carpet and tread often have features that interlock. In a steady-state case (i.e. the robot is operating) traction is conceptually the combination of the surface interactions (like interlocking, adhesion and friction) as well as material properties and failures (is this strain rate sensitive, is the tread flaking off, is cracking of the tread changing the shear forces applied to the floor at a microscopic level, etc.).

In short, I don't really know the true answer, and don't even have a great theory for predicting the low-level interactions. But I think some teams have tried some empirical testing that might shed some light on the issue. Note that it's common in FIRST applications (and I suspect elsewhere) to simply treat all these complications as part of the coefficient of friction. I've seen or used values of 0.9 to 1.3 with good results (for both tread patterns). [Edit: Brecoflex has a summary of "friction" tests for robot drivetrains on carpet.]