Different Drivetrains

[Paul Richardson]

Originally Posted by emeraldstorm

Did your bot turn easily with the grippy wheels in the center? Did it move quickly? Did you have to worry about the omnis taking the force of hitting the obstacles?

I feel like I'm missing an obvious design cue with this. My first thought was "they should've put the pneumatic wheels on the ends and the omni in the center so it could absorb the hits while still being able to turn (with the raising and lowering of the front and rear)", then I wondered if rotation would still be feasible with the friction.

When you think about turning scrub (the resistance to turning from friction), it helps to look at the path each wheel will take when the robot pivots, because this determines the friction force you'll see.

To start simple, imagine a Segway rotating in place about its center. If you traced the path of the wheels, they would draw a circle, right? If you looked at the wheels at any point along that circle, they would be pointing tangent to the curve. In other words, in order to follow the curve, the wheel just needs to roll forwards as the Segway rotates, and there is no sideways movement (which would cause friction). So you can see that the center wheels of a robot have almost no contribution to turning friction.

However, if you switch to a 4/6/8 wheel robot and imagine the circle traced by the corner wheels, you see that the wheels are not tangent to that curve. The corner wheels can't just roll along, they have to slide sideways, too. This is where nearly all turning resistance comes from.

[emeraldstorm]

Originally Posted by Paul Richardson

When you think about turning scrub (the resistance to turning from friction), it helps to look at the path each wheel will take when the robot pivots, because this determines the friction force you'll see.

To start simple, imagine a Segway rotating in place about its center. If you traced the path of the wheels, they would draw a circle, right? If you looked at the wheels at any point along that circle, they would be pointing tangent to the curve. In other words, in order to follow the curve, the wheel just needs to roll forwards as the Segway rotates, and there is no sideways movement (which would cause friction). So you can see that the center wheels of a robot have almost no contribution to turning friction.

However, if you switch to a 4/6/8 wheel robot and imagine the circle traced by the corner wheels, you see that the wheels are not tangent to that curve. The corner wheels can't just roll along, they have to slide sideways, too. This is where nearly all turning resistance comes from.

That's some interesting information. On my way to redesigning my concept drivetrain due to this; Thank you! (is it ok to mention that I repped?)

[Thayer McCollum]

We built a pretty interesting drive-train this year, it was a 8 wheel, drop center, articulated tank drive. The articulation was what made it really interesting. One side of the drive-train (4 wheels-2 CIMs) was solidly fixed to the frame, but the other side was on a pivot underneath the frame. This meant that the sides of our drive-train could be in different planes at the same time. This was really effective at getting over the ramparts specifically because we could drive right up the middle and have both sides of our drive-train flat against both slopes of the ramparts.

https://www.thebluealliance.com/match/2016code_qm3

You can see in this video our robot really bounced back and forth, this was because it would rock on the pivot point due to it not being perfectly balanced.

[happyWobot]

We went with a 35" long x 24" wide perimeter with 4 wheels, 12.5 inch pneumatic configured with #35 chain and two AM Tougbox Minis at 12.75:1 with all wheels dead axle and using Tank Drive as the logic. Wheelbase was about 20 inches axle to axle with a 7 inch separation between tires. This might seem fairly straightforward but here is what we did different.

Chassis floor was 7 inches off the ground to mount the wheel axle below the chassis and bumper was 6 inches off the ground to clear the rock wall. However, while crossing most defenses was silly easy, turning was a problem with significant shudder. The wheels were plastic frame and would bend on turning while the tire shell had knobbies with a lot of grip. The wheel frames would bend so much in a turn that significant energy would get released when the tire knobbies released from the carpet causing the bot to literally bounce several inches off the floor on carpet.

So we tried something different, we dropped the front left and back right wheels 1/8 inch simulating a drop wheel config but on the diagonal. This was highly effective in taking a lot of the shudder out. The diagonal list was barely noticeable and proved not an issue in driving straight. But since the wheels were pneumatic and flexed enough to still maintain some contact with all 4 wheels, the bot still shuddered some, just not violently anymore. We found a slick black duct tape and covered the rear wheels to permit them to drift. Turning was no longer a problem and the config was ridiculously effective at crossing defenses permitting us to avoid the Evel Knievel breaching method.

The diagonal wheel drop and tape cover worked remarkably well even on wheels that have a lot of surface contact due to their diameter and any rocking action was practically unnoticeable. I spent significant time trying to find some reference to someone else having done it before to see if there was any insight on the best wheelbase separation and drop height. Could find no evidence that it had been tried before. I joked a couple times that we should call it the East Coast drive.