Hey guys. Someone had an idea for a drive train on our team that seems a little funky but I can’t put my finger on why. What do you guys think are the advantages and disadvantages of this drivetrain:
Two Large wheels in the back and two perpendicular omni wheels in the front for a type of swerve drive.
| |
Kind of like that.
Are the back wheels swerve?
I don’t think its an omni system you are describing.
It seems like it probably drive like a shopping cart? I feel that would be the case. Of course I’m just guessing, but I feel like the front of your robot would float a lot.
It would be very hard to drive straight unless everything on your back wheels run perfect (proper weight distribution, power distribution, etc.)
Might be too finicky for its advantages.
Just my 2 cents though
Thanks T3_1565. I hadn’t thought of that.
At Ether, I thought that it’s a weak system that can at least swerve; why do you think it can’t swerve.
I believe you and Ether are referring to different definitions of swerve. The standard definition of swerve in FRC is a drivetrain where each wheel can swivel to allow the robot to move in any direction (example). I think you mean that the robot can pivot around one end, which it probably can do to some extent.
It can’t drive in a straight line in any arbitrary robot-centric direction like a swerve can.
It can’t rotate while driving in a straight line like a swerve can.
It can’t pivot around any arbitrary point like a swerve can.
Can you describe why it cannot do those?
If the omni wheels are moving orthogonal to the two wheels in the back, won’t it be able to rotate while driving and pivot an arbitrary point?
Thanks.
Rotating the robot while driving requires rotating the back wheels. Rotating the back wheels changes the robot’s direction of travel.
Note that I didn’t say it couldn’t rotate while driving. I said it couldn’t rotate while driving in a straight line (like a swerve, mec, or omni can do).
Of course it can turn while driving, but so can a drop-center 6WD skid steer.
and pivot an arbitrary point?
It can’t pivot around either of the front wheels, for example. If you think it can, describe what you think the necessary speed of each of the four wheels would need to be to accomplish that.
Switching to a drive configuration that looks something like:
| -
- |
(all four wheels are omniwheels)
…would remedy these problems while being roughly as easy to implement as your proposed configuration. Team 341 did this in 2007 and 2008, and Team 1218 did it in 2010 (I think?).
Inverse kinematics:
*
*
S1 = STR + (L/2)*Omega
S2 = FWD + (W/2)*Omega
S3 = STR - (L/2)*Omega
S4 = FWD - (W/2)*Omega
... then normalize if any wheel exceeds max speed
L = wheelbase, ft
W = trackwidth, ft
FWD = forward, fps
STR = strafe right, fps
Omega = clockwise rotation rate around center, rad/sec
S1 front right wheel linear speed, fps
S2 front left wheel linear speed, fps
S3 rear left wheel linear speed, fps
S4 rear right wheel linear speed, fps
Ether, since this a relatively novel drive that lacks the wealth of information available for other drives, do you think you could list some disadvantages of a drive like this (especially compared to comparable drive trains like slide, mecanum and skid-steer)? I’m specifically interested also in the speed and rotational limitations/capabilities of a drive like this. Also ability to drive straight and the effect of CoMon driving.
Based on CMP footage, it looks like the they used some sort of more standard 4WD in 2010. May have been a different year or a pre-CMP design.
I’ve never built, driven, or even seen a drive like this in action, so take the following speculations with a grain of salt.
Disadvantages
It’s not going to be as easy to drive in a straight line as skidsteer or swerve.
It may require closed-loop speed control of the wheels to get stable and predictable performance.
Traction is poor compared to swerve and skidsteer.
Available force is not equal in all directions, and is less than swerve.
Advantages
It has (theoretically) all the degrees of freedom that a mec or swerve has. (Slide drive does not).
It seems to be rather simple to build and code.
It’s probably lighter that mec and swerve.
It’s probably less expensive than mec and swerve.
(especially compared to comparable drive trains like slide, mecanum and skid-steer)?
I don’t consider slide drive to be comparable to mec and swerve. It does not have 3 degrees of freedom.
and the effect of CoMon driving
What is “CoMon” driving?
*
*
Whoops, that should say CoM as in center of mass. Ie, theoretically, would weight distribution affect this robot’s ability to turn or drive properly?
The kind of drive train Jared is talking about is all over FTC. I’d say 100’s of teams use it every year and it works fine. Great maneuverability, but with only 2 wheels in each direction, not too fast. But that’s where prioritizing comes in. It basically behaves like a low power swerve.
To the OP:
We prototyped the sort of drivetrain (two omnis strafing in front, two tanking in back)
It was horrendous, even with loads distributed evenly. It quite simply did not handle. Good thing we prototyped; we were going to go with that system before that test failed…
Neither have I built or tested this drive system but I am not as down on it as Ether. The first proposed config is flawed but the second one potentially could work, depending on your goals.
The biggest issue I see is that traction is a precious commodity and half of your weight (proportional to traction) is on “undriven” (in the direction you wish to travel) wheels. That said, the same is true of many of the 6 and 8 wheel H-drive configurations I have seen proposed this year. My crude analysis is that it will perform worse than holonomic (mecanum or killough), much worse than 8 motor swerve and comparable to H-drive.
BTW: If you plan on carrying totes “inside” your wheel base, the 90deg wheel may limit your throat size and be a show stopper.