What are everyone’s opinions on what could be an effective drive base for this year and why do you think so?
I had a thought of a swerve drive with treaded wheels. The treaded wheels could provide the necessary traction to move over the BUMPS but the swerve drive provides extreme mobility.
I agree. Crab (swerve) drives will most definitely be effective this year.
Mecanum drives will also be effective, because they are both able to get up the bumps and move with extreme maneuverability and moderate speed. The one downfall is that they have to be dead perpendicular to the ramp, or they will just drive sideways along it.
I think the best drive trains will be those built specifically for this game, such as a drive like this one:
http://www.youtube.com/watch?v=xby0SOphLUg
I keep thinking tank treads would be a good idea. The bumps would be no problem whatsoever.
I tried this year to think about not using my favourite drivetrain (our linkage). But the more and more I thought about it, the more and more I think that our system would work out well. Basically a swerve system with a little less mobilty and less complexity should be well and capable of climbing bumps in at least one or both set ups.
a graduated member of my team suggested using belts that have been flipped inside out to use as tank treds
Some things to consider if you want a swerve drive. Will it high center? How will it climb? Can it take the shock load of falling off the bump? Do you have the resources to ensure this works?
And unfortunately, Simswerve looks somewhat impossible this year. 
Design idea number 40 billion killed by bumpers.
I’d like to encourage teams to draw lots of diagrams, mock up rolling chassis, and think about the assumptions you are making before designing a “drive base”.
Basics
4WD skid - simple, effective, stable climber. If it’s designed to stand a chance in a pushing contest, it won’t be very maneuverable.
6WD skid / tank tread - a bit heavier, more moving parts, NOT a very stable climber, much more maneuverable, still good at pushing (Usually the best “compromise” chassis, but not this year).
Omni / Holonomic - very versatile, simple, maneuverable, but you’re on roller skates, and can’t climb well (or at all).
Crab / Swerve - similar to above but with superior traction, superior handling, more complex, many moving parts, not likely to be a good climber. Does not (usually) rotate the chassis efficiently which could lead to more complex (multi-sided) game mechanisms.
I think the best drive systems will be some sort of hybrid between two of the basics. I would NOT use a tread or a crab to climb… good luck to those who try it! (And do you really need to climb anyway?) If you do, you might want to bring a big tool box to the competitions!
Enjoy!
Why would a tank tread design do poorly while climbing those bumps? I’ve heard the exact opposite: that tanks are great at it, and that they almost never break. I’m on an iPhone, so I can’t link to YouTube, but search up the Ripsaw MS1 made by Howe and Howe Tech (yeah, the ones from discovery channel).
Drive system mobility and ball control are going to be the two most important aspects of robots this year, and likely the two most over looked in the design process.
The advice about checking assumptions and making design drawings and models is excellent, and there is a good list of drive concepts to consider here, as well.
However I would strongly encourage teams to check the assumptions made in the “basics” section of this post for themselves by looking at successful ramp climbers from the Aim High game, and step climbers from the First Frenzy game. Some of what is listed there, although no doubt well-intentioned, doesn’t reflect my personal observations of drive train abilities over the past six years.
Jason
You beat me to this post.
Never rule out unconventional drivetrains that would seem like a waste on a flat field either. Maybe it’s not so good to limit your options to the 5 drivetrains on that list. I don’t know.
I would completely rule out swerve drive if you’ve never built one before. Knowing your limits is key.
What is the reasoning behind not using a crab to climb?
Your reasoning behind many of these suggestions is very flawed. Your references to 6wd robots that fail to climb the ramp are inaccurate generalizations. I have yet to see a ramp that a well designed and well thought out 6wd with very good CG cannot overcome.
Also, this may be the year we see a 6 wheel swerve system. Although I would never doubt someone like WildStang designing a 4 wheel system that can overcome the bump this year.
Tracks. Finally we have a year where tracks are useful again! As for the weight concern, how does 18lbs for simple track modules sound? 38 for simple modules with gearboxes and motors?
I like this year.
When I first started analyzing the game I figured that a basic 6 WD robot with 8" wheels could easily climb the bumps. In this configuration there is no bottoming out issues, and the CG can be kept quite low.
However, we discovered a critical flaw when we tested this.
When a 6 WD robot climbs a ramp*, the center wheel acts as a pivot as the robot climbs over the crest. This results in half the robot being lifted into the air :eek: Highly unstable!
Take a look at the demo bots in the game vid (The non-animated ones of course!). They all have FOUR wheels. This is a much more stable configuration b/c all four wheels remain in contact as the robot climbs. There are no sudden rocking movements forwards or back.
Of course, as Dean himself pointed out, 4 WD robots don’t turn very well… 
*or a tracked drive
False. There is indeed a 6 wheeled robot.
Checking your assumptions at the door is definitely important. For instance, the Aim High ramp was only 30 degrees to a nice flat with a wall behind it. The Bump (caps make it scarier) is 45 degrees up, to a flat with a 45 degree faceplant just waiting for the unwary or timid driver.
Before picking a drivetrain, you should consider how it’s going to rest on the slope, what that implies for your CoM, and what a sudden stop will do to you on the down slope. Or a sudden start on the upslope. You may want to mock up a bump and throw a drivetrain at it to see the effects momentum can have on your robot. I can tell you that a robot with more than 3 contact points per side should worry about having a CoM higher than 12". The case for a 6 wheeler is going to depend on the wheel sizes and separations, though it should be somewhat better.
EDIT: As Martin Taylor pointed out, numerous contact points on a side means you significantly raise your CoM before your robot tips onto the flat of the bump. Which would be survivable if the flat was, say, 24" long. But a 12" flat isn’t a large landing pad. If you have a CoM over 12" in the center of your bot, then once you tip over to the flat of the bump, your CoM is already past the far side of the flat. Which means you just keep tipping forwards. Which means you’re going to faceplant on the far side of the bump. Moving your CoM rearward helps, but not as much as you’d think. Accelerating off the backside of the bump will probably spare you the faceplant at the price of an even rougher landing.
Something I don’t think has been mentioned yet, but is important:
At the kickoff, after the Game Animation, Dean and Woodie demonstrated the difference between using the “slick” wheels and “stick” wheels when it came to the vision tracking.
If we use the “slick” wheels, the robot will have an easy time sliding around lining up with the vision target. However, it pretty much can’t make it over the bump (gets stuck at the top)
If we use “stick” wheels, we can make it right over the bump, but it can’t move sideways to line up with the target.
So if we go with treads, what kind of sideways motion is there to be expected? Sure, treads can go right up and over the bump, but you’d have to resort to tank-drive driving to line up with anything. Not exactly an effective solution, at least in my mind.
I’ve had no problem with tank-drive steering in regards to vision tracking in 2006 or 2007.
I just wanted to point out that the actual grade of the bump is approximately 38.7 degrees (or something to that effect) using simple trigonometry based on the height and width provided by FIRST.