Why swerve is better than Holonomic... HELP

Ok for two years now i have been designing a swerve drive for my team and we as a team finally think it is time for our team to design a omni directional drive train (HOORAH). Before now the team was all for swerve drive. However, some members/mentors changed their minds (due to a vex holonomic a student built) feel that a Holonomic drive is the better way for us to go. Manufacturing and programming of these drive trains are not a problem but because of the simplicity of the Holonomic drivetrain these members/ mentors feel that tradeoffs of the Holonomic drive train will be offset by the time we save in the making and programming of it. As the most experienced student left on the team (and head of design) i feel that we will be better served spending our time on the swerve drive but I am at a disadvantage because i can’t really “prototype” the swerve (like we did with the vex holonomic) without buying all the parts and actually building it. Due to the majority of these students/mentors being new to FRC i feel they don’t understand how much the reduction in pushing power and other weakness of holonomic drive hurts a Robots performance when compared to a swerve drive. Can anyone who has experience driving a holonomic, playing against a holonomic, or has an opionon on the subject give a testimonial about how swerve is better than Holonomic and the making of a swerve drive is the way our team should go.

Thank you very much for your help

Please note: Yes we have built our swerve drive (partially) before so please no warnings about how if you are designing it now it is to late etc…

If you have 2 VEX sets, give 2 groups 1 day for the following challenge: Build a prototype drivetrain. 1 group gets holonomic. 1 group gets swerve. Use standard small green wheels for each drive.

Then, at the end of the day, the head-to-head: pushing, obstacle course, and a sumo match against each other. (Stay in X area. First out, either through pushing or through dodging too much, loses. Best 2 out of 3 or however you want to do it.)

This will give a scale comparison of the two. And yes, it is possible to build a swerve in VEX.

330 did something similar back in 2005 with a mecanum vs. a 6WD “drop”. A full-scale kitbot mecanum drive went up against our 4WD 2003 robot in a match situation–and lost.

The match will do several things: It will expose the strengths and weaknesses of both systems. It will give practice with both. If you do it right, you’ve even got the code for both. And, there will be one of two final results: Either you’ll see that you’re wrong, or they’ll see that swerve is superior.

Edit: We also want to go over the bump. I also feel that the Omni wheels would not have enough traction to be able to go over the bump

I wonder what a key element in that obstacle course I mentioned earlier might be…:rolleyes: Shouldn’t be too hard to build one to about 1/3 scale.

I am having a similar problem, only it’s the mentors wanting to use a 6WD 2 gearbox design and a handful of students wanting to use holonomic, and the 6WD prototype was built by a mentor. What ever happened to student decisions?

Omni wheels still have quite a bit of traction on carpet in their driven direction.

Video of Omni wheel robot up the bump

An important thing to consider here, and it speaks volumes about your team’s personality, is the following: Is it an adult decision, a student decision, or a team decision?

Build a vex holonomic and a vex 6 wheel and have the Vex 6 wheel play defense on it. Should sell your team pretty quickly on not holonomic.

Have you built, tested, rammed, destroyed, rebuilt, retested a swerve drive? If you have, then great. The only reason (if swerve is desired) you’d want to run holonomic over it is weight savings, simplicity, durability, at the expense of all traction versus defensive robots. If you haven’t, your team’s got a point in that swerve is deceptively hard.

If you want, send them videos of 1714’s 2008 robot, which is holonomic. (Just don’t tell them there were programming bugs :))

Does holonomic really lose all that much traction? What if you use braking on the Jaguars? It seems that it wouldn’t be too bad, the only thing hindering you is the hard plastic of the wheels.

First, a picky technical note:
Holonomic is a general term used to describe a system that can be controlled in all of its degrees of freedom [1]](http://en.wikipedia.org/wiki/Holonomic#Robotics). For robot drivetrains, this implies the ability to move in two dimensions. A swerve drive, mecanum drive (jester drive, airtrax), and omnidrive (kiwi drive) are all examples of holonomic drivetrains. They allow movement in both dimensions.
I’m going to assume that my “holonimic drive” you are referring to mecanum drive.

Second, there are major advantages and disadvantages to both swerve drive and mecanum drive. I’m not trying to be rude, but please don’t come to this community to enlist support for your side. Take a look around these forums; you’ll see that the relative merits of these drivetrains have been discussed numerous times before.

I know you honestly believe that swerve drive is the right way to go. In terms of field capability, you’re probably right. But there are a host of other factors, most notably complexity, weight, and manufacturing time, which all affect the decision. And in the end, those factors are actually more important.
There will be great swerve drive robots this year. But there will also be some great mecanum drive robots, and some great omni drive robots. There will even be some great robots that use the kitbot drive train. An awesome drive train might help, but solid ball handling and hanging capability will determine how you do. Don’t let yourself think that your drivetrain choice is going to doom your season.

Feel? The only way to know is to try it. Honestly, I’d be interested to see what happens - 2359 is asking many of the same questions.

Yes. Even if you have wheels with as much traction as any other drive system, you inherently lose between 50% and 71% of your power. Take a look at this paper for an explanation why.

If you really want to be clever, just ask your team how many mecanum robots have been on Einstein.

I think most people mean 4 omnis, two perpendicular and two parallel (Killough), when they say holonomic.

Also, you would lose the 50%-71% of power, but not traction, so people pushing you shouldn’t be too harsh if your wheels are locked. Given you won’t have as much power pushing back.

For my presentation on drive trains that I do for our technical conference in Oregon, FIRSTFare, I count up the types of drive trains used by the finalists and winners of the Championship divisions. 2009 was a weird game so setting that aside, 2008 is the last year with useful data. Here are the totals:

14 Six Wheel
2 Six Wheel with omnis
2 Four wheel with omnis
2 Mecanum
2 Crab Drive
1 Four wheel rack and pinion

This was just from the info I could glean from looking at pictures of these robots so I can’t claim this to be 100% right but it’s close.

Keep in mind that was a very different game than this year so I’m not sure how much weight to put on this. It was also only the second year that Mecanum was widely available from AM and the first year they had 6" Mecanum wheels.

I hesitate to say you lose traction, because your wheels may still have the same grip on the carpet. But even if you lock your wheels, you only have a fraction of your “pushing resistance force”. Unless you have a way to physically lock or articulate the rollers as describe in the paper I linked above, up to half of your traction is wasted on rollers that simply spin.

Take a square robot with one omniwheel (rollers perpendicular to the drive direction) on each side. To move forward, two wheels are doing the work, and two wheels are spinning their rollers and contributing nothing. The scenario is identical if we brake our wheels and another robot tries to push us. Two wheels will give us traction on the carpet, but the other two give us nothing.

If the robot moves diagonally, all four wheels move, but all of the rollers spin some. This results in a 1/(sqrt(2)) thing, which is where I get 71% power/traction.

Sorry, I really worded that wrong. I made it sound like I’m angry because the mentors won’t let me do what I want to do. But that’s not it. I just feel that if the mentors don’t support it, it won’t happen, no matter what. A few students honestly believe holonomic drive will work (that’s not to say all though), but the mentors don’t want to, and so there’s no use in pushing for it because their word is law. You’re VERY right, it should be a joint decision.

Thanks for the clarification, that makes sense. I still think the tradeoff is worth it though.

If they really want to use omni’s, tell them to use linkage drive. Yet something tells me that we’re not getting the full picture about how the mentors truly feel the swerve drive contributed to the robot in the previous years. Perhaps they don’t want to repeat the stress of creating and maintaining the robot, or perhaps they want more time to work on the manipulators while still maintaining certain advantages of strafing.

As for the ramp, you want all 4 wheels contribute to climbing, even if they’re pointed 45 degrees off of straight forward. Unless you have a GOOD suspension one of the four wheels WILL lift off of the ground if the robot isn’t lined up straight. Any 1/3 scale model, even if built out of toothpicks, will tell you that. So omni drive trains should be ok with going up and over the ramps should they choose to ‘gun it’. Yet soon the mentors will realize that what goes up quickly will also come down quickly. Now either that can be a tradeoff you live with or it’s a design consideration that affects a decision.

All engineering is tradeoffs … and understanding those tradeoffs and matching the best options to the strategy you choose is where the real engineering comes in.

No drivetrain is the best. Each has it’s advantages and disadvantages and these must be weighed against your strategy, your teams capabilities, and your understanding of how the game will play out.

In the end, the best drivetrain/manipulator/kicker/hanger/circular-duck will be the one that most closely assists your chosen strategy.

Our team built an omni drive in 2008, and it only worked ok. We built a vex omni drive, programmed it, and it worked great. Vex can be a good tool to prototype something, but the FRC robot did not drive any where as well as the vex robot did. Could more complex programming have helped? Possibility, but in our experience, the omni drive did not work well.

The traction will not likely be the challenge for a holonomic drive base getting over the bump. Unless you have some sort of suspension, it will be virtually impossible to maintain the wheel contact and proper weight distribution to the wheels in order for it to drive properly.
As you go up the bump, one or more wheels will likely be lifted completely off the ground (unless you’re using 3-wheel base that’s perfectly aligned with the bump), and thus that wheel can no longer contribute to any driving you’re doing (which will likely change the direction you’re driving, which will likely cause you to no longer climb the bump). Additionally, as your robot is angled, the weight distribution to each wheel will change, also changing the resulting normal force on each wheel (which could also cause a change in direction).

There is simply no substitute for prototyping. If you want to know if a drive can climb, defend, push, turn or move sideways effectively, the only way is to make one and put it to your own test. By now many teams have done just that and have played the game, even if only with students portraying robots. Most have come to the conclusion that some pushing will take place, some climbing over the bump will be needed and accurate movement and position of the robot for kicking is a must if you want to score. If you prototype you will know if your robot will slide sideways off the bump, if it can be pushed easily or if the drive system will work at all.
I will answer again as I have so many times before. We do not decide on a drive type until we brainstorm, prototype and test. Now one should.