crab drive vs. mecanum drive system

[EricH]

Originally Posted by superbotman

My team is debating on the drive system we want to use and I thought I would ask what has worked well for other teams.

1. Crab drive works best

2. Mecanum drive with each wheel having its own motor wins.

3. Use mecanum drive with one motor per side.

4. don't do any and use some other drive system

Crab is tough to build, though there are now swerve modules available. Needs a customized frame and coding. Also note that there are two types: coaxial and non-coaxial, each of which has its own unique challenges.

Mecanum drive is simpler in the frame (KOP frame works), but you need to make sure the wheels are tuned right. I'm going to have to disagree with daltore on the diagonal, though--we never saw anything when we tested our set back in 2005. A MAJOR pro here is that if it doesn't work, you grab 4 wheels from the parts box and slap them in, then remove the sideways code, and you have 4WD.

Mecanum with only one motor per side is impossible. Omnidrive (4-wheel) gives one motor per wheel, one wheel per side, or something of that sort.

Many teams use a 6WD with either a dropped center wheel or omnis on one end or the other. That's the simplest, and the KOP frame is set up to handle it already.

At this point, though, you're about out of time to prototype anything other than a 4WD or 6WD, unless you get an order in to AM immediately.

There is one other consideration, however:
We don't know the game yet, and you're choosing a drive to use for the game. Stop. Think about it. You could be investing hundreds of dollars in a system that won't fit the game at all. Or something like happened this year, where mecanum wheels and omni wheels couldn't be used.

If you go ahead and try any of these systems, do it as a prototype in the offseason (the month left of it, or the next one) to get the hang of it, then build it for the competition season following that when you actually know what you're doing. Can you do it during the build season? Probably. Can you do much else? Probably not.

I know that when 330 tried their set of mecanums during build season, we already had a couple of years of small-scale testing under our belts, and a full-size backup that we could literally swap in a few minutes. (I'm talking VEX scale here--the EDUbot was great for that!) We had code, we had wheels made after Kickoff, we had the Kit frame rigged for mecanum pretty quickly, and we had a 6WD drop center on the competition frame. Last week of build, we had a decision meeting, did some final testing, and the mecanums lost to the 6WD drop center. If we ever try mecanums again, we'll be almost ready to go. That's why you prototype. But if you lock into a drive now, you might not be able to change when you realize that being able to dodge around opponents isn't going to help you this year, or that you really want that extra maneuverability, or this, or that. Seriously, don't lock in now.

[Chris is me]

Oh, and my two cents: I've personally never seen an FRC game where, in my opinion, a mecanum drive's advantages outweighed its disadvantages compared to a 6 wheel system

[AdamHeard]

Originally Posted by Chris is me

Oh, and my two cents: I've personally never seen an FRC game where, in my opinion, a mecanum drive's drawbacks didn't outweigh its benefits over a 6 wheel.

Definitely Debatable.

Originally Posted by AdamHeard

Definitely Debatable.

You're debating that he hasn't seen one where it's beneficial?

[Chris is me]

Originally Posted by CraigHickman

You're debating that he hasn't seen one where it's beneficial?

I think he's debating my opinion, not whether or not I have my opinion.

2001 doesn't count.

(By the way, my post's wording SUCKS and was EXACTLY the opposite of what I meant, sorry. I meant 6wd > mecanum, not other way around.)

[Vikesrock]

something to keep in mind for those responding specifically to the OP, team 2169 had crab drive on their 2009 robot.

I think it is important to wait for the game before making a decision between the two drive systems. I also think it would be wise to keep simpler alternatives on the table.

After analyzing the game and deciding on what role you want your robot to play you can list out what drive capabilities you need to best accomplish that role. This should make it much easier to determine what drive system you want to go with.

I saw at least a few swerve drives in 2009 that did not appear to contribute to the team's strategy. I would make sure you have a reason before you expend the effort, weight and increased complexity to build a swerve drive.

[dtengineering]

The decision to build an omni-directional drive train... however you do it... doesn't have to be based solely upon what is best for winning the game.

It should be based on what is best for your team.

Rarely have I ever seen a truly dominant robot dominate because it has an omni-direction drive system. In fact, given our experience with a mecanum system, the additional degree of freedom actually makes the robot slightly more difficult to drive and it takes longer for the driver to learn to use the ability productively... until then it just makes it more complex to drive.

However, if you have a dedicated, well-balanced team with solid machining and programming abilities and are looking for a challenge... especially if you haven't built an omni-drive robot yet... then consider building one just because it is something cool and technically challenging that you have yet to do!

Omni-bots are really, really cool whatever method you use to achieve the goal.

Jason

[AdamHeard]

Originally Posted by Chris is me

I think he's debating my opinion, not whether or not I have my opinion.

2001 doesn't count.

(By the way, my post's wording SUCKS and was EXACTLY the opposite of what I meant, sorry. I meant 6wd > mecanum, not other way around.)

My apologies then, 6WD is the way to go usually.

[sgreco]

I'll be honest, I have nothing quantitative to share with you, but...A really good 6 wheel skid is hard to be beat. It's lighter than a swerve and if you look at the teams that do it really well, their performance really isn't that much worse if worse at all from a good swerve. Keep the driver in mind too, different drivers prefer different systems. Skid steer requires basic code, successful swerves require much more complex code, especially if you want it to be intuitive.

[JesseK]

I shudder at the thought of Mecanum with only 2 CIMs. You might as well disable trying to rotate while moving in any transverse direction. You also might as well disable the diagonal movements. The wheels simple eat up too much torque due to the 45-degree roller offset.

Simply put, either drive train gives a good driver very similar capabilities. Crab gives the driver an edge in finely tuned movements, yet Mecanum may give the driver more practice since it's much simpler to build.

With Mecanum, square up on another robot before pushing it. Programming is relatively simple. The Toughbox Nano's make a Mecanum drive train easily constructable since 4 nano's with cantilevered Mecanum wheels should be able to support the entire weight of the robot without issue. Seriously, If AM redesigned the toughbox with a steel box extrusion like what the nano has, they would probably be strong enough for military-grade and/or volatile environment (think chernoble cleanup) robots.

With Crab, your driver needs to dictate the driving style, which then dictates how the 4 wheels are steered (2-2 sides, 2-2 front/back, or all together) as well as how the robot is programmed. This driver would definitely be more focused on the robot during a match and less focused on what's going on around him/her.

I'll be honest, I have nothing quantitative to share with you, but...A really good 6 wheel skid is hard to be beat. It's lighter than a swerve and if you look at the teams that do it really well, their performance really isn't that much worse if worse at all from a good swerve. Keep the driver in mind too, different drivers prefer different systems. Skid steer requires basic code, successful swerves require much more complex code, especially if you want it to be intuitive.

It's all about how you use the drive train. A 6-wheel drive train has shortcomings when attempting maneuvers, especially in autonomous. For great examples of what I'm talking about, look at Wildstang's 2007 and 2009 robots. Both autonomous routines used maneuvers with swerve that would have probably been much less accurate (statistically) than had the maneuvers been performed with skid -- especially in 2009.

[Brandon Holley]

I'll just give my real quick 2 cents because I am compiling a post which contains all of the information from my senior design project at Northeastern University.

My senior design project was the investigation of the most effective drive system one could use in FIRST, followed by the optimization of such a drive and then constructing a prototype. I went through a lot of decision making processes to determine what was most effective. When it came down to it, the argument was exactly what is being argued right now...swerve vs. 6wd. Some of the decisions I made to determine between the two were based on opinion, but this is inherent with this kind of decision. I ended up designing, analyzing, optimizing, building, assembling and testing a swerve system.

The argument between the two was settled for me based on the fact that I feel the only reason people widely consider 6wd more reliable is because they are simply done much more frequently than swerve/crab systems. Yes there are more moving parts in a swerve system, inherently making the system more prone to failure. However, with more evolution of swerve design the systems can become just as reliable as 6wd systems today. Which is why I chose to optimize that system. A new post will be up in a few days which contains my report (>100 pages) and my presentation (~40 detailed slides) which explains what I did and the findings I had.


Brando

[Chief Samwize]

I believe it should depend on the game. Personally, I dont much care for mecanum drive trains however they can be useful in some games. For example, in 2006 I cant remember the team number but their was a team that climbed onto the ramp and then using the mecanum drive would move all the way to one end of the ramp to allow room for their partners to get on the ramp as well.

-Sam

[Chris is me]

Originally Posted by JesseK

I shudder at the thought of Mecanum with only 2 CIMs. You might as well disable trying to rotate while moving in any transverse direction. You also might as well disable the diagonal movements. The wheels simple eat up too much torque due to the 45-degree roller offset.

I think the bigger issue with 2 CIM mecanum isn't torque but that you can't independently control 4 wheels with 2 motors. Well, with any amount of reasonability.

[Raul]

Originally Posted by JesseK

...
It's all about how you use the drive train. A 6-wheel drive train has shortcomings when attempting maneuvers, especially in autonomous. For great examples of what I'm talking about, look at Wildstang's 2007 and 2009 robots. Both autonomous routines used maneuvers with swerve that would have probably been much less accurate (statistically) than had the maneuvers been performed with skid -- especially in 2009.

I have a better example: I could easily argue that what we were able to do in 2003 in autonomous was both, extremely advantageous, and almost impossible to do without Crab steering.

[Kingofl337]

Team 40 has tried most of the drive combinations mentioned here over the last few years.

2005 - 6wd Solid Center, front/rear omnis
Robot was good a could maneuver well, be wasn't very repeatable in in autonomous. Our implementation caused the robot to rock on the center wheel. I was also difficult to drive, the robot wouldn't drive strait to save it's life.

2006 - Mecanum w/3 Speed Dewalts
Robot could go in any direction, the drivers did seem to complain about difficulty driving it. It worked great for auto tracking and shooting in autonomous. On the other hand the robot had zero pushing power, and could be easily knocked off coarse in auto or shoved in a corner during operator control. It would take a lot of convincing for us to go back to this drive train.

2007 - 6wd Drop center, 2 speed servo shift by AndyMark
The robot drove well, it was easy to program, and we used a turret in place of maneuverability. While the turret was nice, we missed the ease of navigation. Also it rocked on the center wheel which would cause the drivers to readjust the tube.

2008 - 4wd Swerve, Motors In Pods, Pods Grouped F/R for Steering
Based on the 2008 Control System, this was extremely difficult to program and still have enough compute power to control all the other devices. Autonomous was very easy to program with "robocoach" and allowed the robot to strafe at will. Driving the robot was like driving car and was extremely natural, and through software the robot could enter crab mode and strafe in any direction. Other then a few broken POTS this drive was extremely reliable. We are definitely sold on this drive system, if we have 6 motors to dedicate to the cause.

2009 - 3wd Crab Drive
We liked the swerve so much in 2008 we did it again in 2009. Though we ran out of motors and space so we did a tri-cycle drive and linked the steering together and used a single motor to steer instead of two. Programming in the 2009 system with C++ was very simple with the included PID functions. With the surface being so slick the robot could still rotate through software tricks. This allowed us to have a extremely successful autonomous. The drivers did struggle a bit more this year, as driving directionally and slow rotation caused some issues.