There’s another answer that I haven’t seen yet. So I’ll give it.
It’s because you actually lose pushing force if you have unpowered wheels on the ground. And you’re pretty much guaranteed–with a 6WD–to have unpowered on the ground.
Here’s the physics behind that.
First, we assume that Traction is mu* Fn, for any given wheel. Fn is the normal force of the robot, divided by the wheels in contact with the floor; mu is the frictional force between the wheel and the carpet. (Also include any manipulators–but that’s not important here unless you’re driving a manipulator into the carpet.)*
Next, we assume that for a 6WD robot with a dropped center wheel, no more than 4 wheels are on the floor at any given instant, and possibly only 2. Also assume that the robot weight is 150 lb.
Now, a few case studies.
- 2WD, 2 wheels on the ground
in this case, each wheel has 75mu lbf of traction. 150mu total.
- 6WD, 2 wheels on the ground. See case 1, but the outer wheels spin with the inner ones.
- 2WD, 4 wheels on the ground.
Each wheel now has 37.5mu lbf of traction. But, with only 2 wheels driving… 75mu lbf total. The rest? Well… that’s taken up by the two non-powered wheels… uh-oh.
- 6WD, 4 wheels on the ground.
Each wheel has 37.5mu lbf of traction… and there’s 4 wheels on the floor. 150mu lbf total traction.
When 4 hits 3, 3 is likely to lose any resulting pushing match just on available pushing force. Oh, and did someone mention that the robot may or may not actually be on the drive wheels?
It’s a couple pounds of belt or chain that can have a pretty large effect on robot performance. Between ability to get unstuck and the theoretical traction boost, as well as a reduction in drivetrain drag and/or scrub…
*For the pedantic/detail-oriented–not all wheels will have the SAME normal force due to things like CG placement at the moment we’re observing, how tipped up the robot is… stuff like that. But for an overall approximation, it’s a reasonable assumption.