Drop-center drivetrains: Why?

[JamesCH95]

Originally Posted by billbo911

Now, if you want to climb a ramp, or go over a bump, then omni's on the corners add a whole new level of difficulty. I'm not talking so much about traction, but the need to hit the incline squarely. If you are out of alignment by only a few degrees then the robot will tend to slide out, or fall back to the level surface. The steeper the incline, the greater this effect will be.

Although we did not use omni wheels this year, we did us mecanums, and we noticed that the robot wouldn't glance off inclines, but rather tend to align with them. There probably is some angle where the robot would tend to glance off rather than self-align, but my experience tells me that this angle is sufficiently wide to easily drive over a bump or up a ramp.

Originally Posted by Chris is me

I really think "just use omnis" can be a pretty poor solution depending on the game challenge. If you have a long arm, a gentle tap moves your game piece several feet away and there's not much resistance to speak of. Really, it's not that hard to do "all of the analysis". I can tell you what the analysis will say right now: The more rigid your chassis, the better your 6WD can perform. Building a rigid chassis doesn't take tons of time and energy. Even if you do want to do careful and thorough analysis for such a drivetrain, it's not exactly a huge resource drain, and depending on the challenge it could result in a much, much better chassis for your needs.

While some games you probably could get away with a chassis with four omnis, in the past I think more often than not you couldn't. Being spun off target in Aim High, knocking you away from your spider leg in Rack 'n Roll, or taking a mid to long range shot in Breakaway are all situations when I'd rather have a bit more resistance to being spun.

Perhaps you did not understand that I was referring to a 6wd chassis with two traction wheels in the center and four omni wheels on the corners. In my experience these have been fairly resistant to being spun and do not rock much, if at all; a good compromise in my opinion.

I was not referring to a chassis stiffness analysis, but rather keeping track of CG location for the purposes of reducing rocking while manipulating a game piece. While these are both straight-forward analyses to perform, not every team has a lab full of CAD-capable computers and/or trained operators ready to do them.

[Chris is me]

Originally Posted by JamesCH95

Perhaps you did not understand that I was referring to a 6wd chassis with two traction wheels in the center and four omni wheels on the corners. In my experience these have been fairly resistant to being spun and do not rock much, if at all; a good compromise in my opinion.

I was not referring to a chassis stiffness analysis, but rather keeping track of CG location for the purposes of reducing rocking while manipulating a game piece. While these are both straight-forward analyses to perform, not every team has a lab full of CAD-capable computers and/or trained operators ready to do them.

I caught that. 6 wheel 4 omni isn't terrible or anything but it is more easily spun and negates a lot of the advantages of a rocking drivetrain. The CG issue, while relevant, really isn't something you need computer precision to analyze. If you don't want to rock with an arm too much, you can adjust the drop a little, put the CG a bit more to one side fore-aft, or even run an 8 wheel drive if your CG is centered. You should be able to get a good enough estimate with just some napkin sketches of where the "heavy stuff" goes. CAD isn't required to figure it out.

Even if you do find you need more precision than "I think it'd be best if we put the battery... here", CG analysis is always beneficial on a competition robot.

[Cory]

People refer to "rocking" all the time and cite it as a problem with manipulators.

Think about the math here. With a 33" wheelbase (37" long chassis, with 4" wheels), that means you have 16.5" between wheels. With a .1875" drop on the center, you can rock 1/100 of a degree. Obviously that's exaggerated when you have a long arm sticking out, but come on. .010 degrees of rock is nothing. It's hardly noticeable. It doesn't affect precise positioning whatsoever.

[AdamHeard]

Originally Posted by Cory

People refer to "rocking" all the time and cite it as a problem with manipulators.

Think about the math here. With a 33" wheelbase (37" long chassis, with 4" wheels), that means you have 16.5" between wheels. With a .1875" drop on the center, you can rock 1/100 of a degree. Obviously that's exaggerated when you have a long arm sticking out, but come on. .010 degrees of rock is nothing. It's hardly noticeable. It doesn't affect precise positioning whatsoever.

Agreed.

Even with a 10' arm on there, if your manipulator needs to be accurate in placement to 2" or less, you really have a poorly designed manipulator that would be difficult to score with anyway.

[JesseK]

An interesting idea inspired by this thread would be to (gasp!) raise the center wheel on the fly in order to reduce turnability similar to the nonadrive concept.

Hmm.

[AdamHeard]

Originally Posted by JesseK

An interesting idea inspired by this thread would be to (gasp!) raise the center wheel on the fly in order to reduce turnability similar to the nonadrive concept.

Hmm.

For what "benefit"?

Before people start going off and design new for the sake of new, they should evaluate what a well designed full treaded drop center 6wd can do. There is a reason 60 started doing it, 254/968 always does, and we always love to copy it.

[JesseK]

Originally Posted by AdamHeard

For what "benefit"?

Before people start going off and design new for the sake of new, they should evaluate what a well designed full treaded drop center 6wd can do. There is a reason 60 started doing it, 254/968 always does, and we always love to copy it.

If what I've suggested is "new for the sake of new" implementation of 6WD/WCD is "a solution searching for more problems". I made the benefit pretty clear, but perhaps a more in-depth explanation is required. In short, the benefit to raising the middle wheel on the fly is to maintain a level of functionality while removing the derived assumptions realized by a more advanced/coupled design. Indeed, I did not even imply those assumptions, but then again no one ever really talks about them.

Reduce the team in question to an average team who seeks to maintain most of the same capabilities as the 'best' 6WD while also maintaining other requirements they've set forth for their robot. Perhaps 4-6 motors and a COTS shifting transmission on the drive train is deemed a lower priority than having extra power/weight for other robot subsystems for a team. To be honest, this is a very reasonable assumption for any team.

If the extra power/weight in the team's manipulators were to pay off in on-field success, many other teams would come to defend them.

Thus the team would be subject to defense via turning due to its gearing choices and wheel base when it competes against another robot with more power/capability in the drive train. Adding a potential design to raise the middle wheel using simple pneumatics in order to remove the disadvantage may prove to be a superior design for that team's overall robot depending on their time, available resources and funding.

[IKE]

We had a variable drop center chassis this year. In simplest terms, it allowed our middle wheel to raise up while going over the bump (4-5"). This allowed us to have smoother and faster transitions with less impact on the robot/chassis. This was a key component of our strategy as we knew that for many matches we would start in the back and work our way forward and did not want to be limited by the potential tunnel pinch point. Our analysis (check white papers) showed a flat or rock 6x6 would have a considerable amount of problems going over the bump (not impossible, just not pretty).

The variable drop middle wheel also allowed us to flatten the chassis to "hold" directional heading (very important in 2005, 2006 and 2007). As it turns out, this year holding a particular heading was not as essential as those other years. You were either pointed at the goal to take a shot, or you were not. If we had been allowed to store multiple balls, this would ahve been a different story, but it also would have been a very different game.

This variable ride height middle wheel did cause issues with our collector.

[Siri]

Originally Posted by Al Skierkiewicz

We have (on more than one occasion) used a dropped a foot under the front or back of the robot to assist in turning. This raises one set of wheels off the ground.

We did something similar in our non-dropped 6wd (wide) for Lunacy. It was actually a pneumatically actuated wheel that lifted up the back four, which was great for both turning on a dime acting as a break on the regolith. We could even strafe on occasion, though certainly not as well as nonadrive.

Lesson being there are a lot of novel "dropping" options to achieve your desired results--it need not be just the center. (photo)

[JamesCH95]

Originally Posted by Chris is me

I caught that. 6 wheel 4 omni isn't terrible or anything but it is more easily spun and negates a lot of the advantages of a rocking drivetrain. The CG issue, while relevant, really isn't something you need computer precision to analyze. If you don't want to rock with an arm too much, you can adjust the drop a little, put the CG a bit more to one side fore-aft, or even run an 8 wheel drive if your CG is centered. You should be able to get a good enough estimate with just some napkin sketches of where the "heavy stuff" goes. CAD isn't required to figure it out.

I am aware that CG/tipping moments can be roughly estimated with napkin calculations. Chassis stiffness is not analyzed so easily.

We've also done a drive-train with two omni wheels in the middle and traction wheels on four corners these omni+traction 6wd robots, in either configuration, have all been successful.

I'm not trying to invalidate your points, Chris, they are indeed accurate. While the does trade off a little hit-and-spin stability, it also gains turning speed and virtually eliminates jitters from scrubbing the outside wheels and rocking. As with any robot design choice it has to be made in context of the teams strategy in that year's game. I feel that this disclaimer need not be made with every post though.