What type of drive train is the most maneuverable?

[Lil' Lavery]

In terms of pure maneuverability, mecanum is the best, as it can travel in any planar direction, change direction almost instantly, and can conquer inclines easier than a holonomic drive. Holonomic and ball drives can also travel in any planar direction and change direction almost instantly. Swerve drives can also travel in any planar direction, but they recquire time (often very small though) to change the orientation of their wheels. "Crab drives" (a variant of swerve drives where all the wheels are linked together) are the simplest swerve drive, but recquire more power (as the have to overcome more scrub) to move in curves. Afterwards comes the plethora of different scrub (tank/skid) steer systems.
4WD, 6WD, 2WD (and 2WD+caster), and many others all fall into this category. While some may recommend the caster system, I would advise against it. From my personal preferance, while quite maneuvable, the lack of control on the casters can cause more problems that it solves. It significantly reduces your control over your bot, and therefor reduces the amount of accuracy you can drive with. Momentum can often cause the rear end (assuming your casters are in your rear) of your bot to swing further than you want it to. It also makes you a relatively easy target to push (and incredibly easy to rotate out of position).
The "best buy" for maneuverability (in my opinion) would be a 4WD with 2 omni-wheels.

[Andrew Blair]

I definently agree with Lil' Lavery- The best bang for the buck if you're not concerned with lots of defence or complication will be the four wheel setup and omni's out front.

Six wheel is another excellent choice with a lowered center wheel- but it also requires more time, complication, and weight budget. However, it's nearly impossible to be spun sideways, or lose traction on inclines, bumpy ground, etc.


Mechanum and swerve are very impressive and capable designs when the game warrants them, but compared to standard drive-trains, a huge amount of design and fabrication muscle needs to go into them.

[Lil' Lavery]

Originally Posted by Andrew Blair

I definently agree with Lil' Lavery- The best bang for the buck if you're not concerned with lots of defence or complication will be the four wheel setup and omni's out front.

I actually prefer omni's on opposite corners (like 494/70) as opposed to 2 on the same side. That way it makes you harder to be spun by other teams, but still allows you to spin and change direction quickly and accurately.

[Tom Bottiglieri]

Originally Posted by Lil' Lavery

I actually prefer omni's on opposite corners (like 494/70) as opposed to 2 on the same side. That way it makes you harder to be spun by other teams, but still allows you to spin and change direction quickly and accurately.

That is an awesome idea...

[Richard Wallace]

Originally Posted by Tom Bottiglieri

That is an awesome idea...

Yes it is. And effective. Recall this rather intimidating scene?

[Stephen Kowski]

in the idea of a slim budget, the omnis are about $80 for 2.....extra skyways that can easily be set up with the kit drive train are much cheaper. if you have the extra money the 4wd with omnis are a good option, but if you don't have the cash or would rather save it for another area id stay with kit 6wd drive with the lowered middle wheel and I know 25 doesn't lower their middle wheel which is fine, but the kit bot is already setup for it.....

[KenWittlief]

I think the simplicity and low energy requirements of a 2 wheel drive system are being overlooked:

1. with a 2 wheel drive system you dont have to scrub any wheels to turn. It takes very little power to steer

2. instead of using castors, you can use low friction material and have skid plates on the corners. Anyone who has ever pushed a shopping cart knows castors take some force to get them to point in the direction you want them to go. The result is the force required to turn is non-linear until the castor rotates. A skid plate solves this problem, or you could use a ball style castor (its like a big ball bearing that is free to rotate in any direction.

3. You can compensate for the 'being pushed sideways' and 'hard to steer up a ramp' issue very simply, using the yaw rate sensor (gyro chip) that has been in the KOP, and using a simple feedback loop to close the loop on the steering. Then you have the best of both worlds, a robot that will spin on a dime, taking very little energy to do so, and that will use all of its available motor power to keep pointing in the direction the joystick is commanding, no matter who or what (gravity) is trying to make it turn.

[efoote868]

i think we're all over looking the hovercraft... ( )

For a really tight budget, that would probably rule out crab, mecanum, or omni (spending money on wheels and the drivetrain might not be most well spend, depending on the game)

Probably one of the easiest (and cheapest) to impliment would be the 2 wheel / skid plates.
!*caution*! with the gyro, make sure its rated to a high rotation, when we tried to use that with our robot this year, it messed up because it could only do 80 degrees/sec !/*caution*/!

I would say experiment as much as you can in the off-season, build a generic frame, and try different configurations with the wheels, OR browse soap's collection of matches, and take a peak at each of the videos.

I personally can reccomend the swerve system, as 114 had great fun with one of those. The only issue is how to program the thing, and still have it driveable.

As far as most cost effective and still maneuverable, a 6 wheel with the middle wheel lowered is pretty nice.

[Veselin Kolev]

mecanum and holonomic are very maneuverable, but they only have on average at best 70% efficiency. And they're arguably more complicated to program, needs sensors, etc.

swerve is a complicated system of wheel modules and takes time and money to fabricate, but is more efficient (more pushing power) than mecanum/holonomic. Still has a bit of programming to do, and sensors. swerve can also get heavy if weight isnt seriously taken into consideration during design.

6 wheel drive is simpler to fabricate, cheaper, and easy to program. however, you cant strafe sideways. you can turn arguably better than most other drives, but your robot will rock back and forth a little, which isnt always good.

personally, I'd go with 2 speed swerve. all you guys that think its too hard to pull off, too heavy, too expensive, too complicated, or overkill can have their opinions.

[Tom Bottiglieri]

Originally Posted by Veselin Kolev

personally, I'd go with 2 speed swerve.

I guess you want to have your cake and eat it too!

Some words of advice.. Don't try to do too much! If you want to be agile and escape defenders, then do it. If you want to push defenders out of the way, the n do it. Don't try to switch up mid match as it will only cause you to waste valuable time.

The smartest thing you can do is analyze the game when it comes out and devise a strategy to your liking. Base your design goals off of this, and seriously consider what each available setup can add to your overall design.

[efoote868]

Originally Posted by Veselin Kolev

...And they're arguably more complicated to program, needs sensors, etc....

Our programmers this year actually got the code down to 4 lines... and it was just converting joy numbers to good PWM values.... super efficient.

Most programmers that I've met actually say they're easier to program than a 4-wheel / 6 wheel tank drive, or at least the ones that have used them.
Only a very weird configuration would make it harder... like only using 3 wheels

As for making it go relative to the driver, thats where it becomes tricky (all that great vector math). Thats where we used the gyro, and thats how our robot messed up (preseason, the lead programmer then decided that it wasn't worth our time).

We also found out that a gyro is affected by temperature, pressure, etc. greatly; and that it gave out different readings at different times of the day oh well

[Lil' Lavery]

Originally Posted by efoote868

As for making it go relative to the driver, thats where it becomes tricky (all that great vector math).

How would vectors not apply for robot-centric control as well?

[KenWittlief]

Originally Posted by efoote868

...We also found out that a gyro is affected by temperature, pressure, etc. greatly; and that it gave out different readings at different times of the day ...

if that was true then your gyro was defective or there was a bug in your code reading the gryo value. Solid state gyros are highly reliable, the same AD chips used on the robots are used in cars for things like airbags and stability control (steering).

One thing about the gyros, they must be solidly mounted to the chassis of the robot. If the gyro can wiggle and vibrate (if you hold it down with tie wraps, or foam tape) you will get all sorts junk on the output signal.

To be useful the gyro chip must move exactly the same as the robot chassis moves.

[KenWittlief]

Originally Posted by efoote868

...Probably one of the easiest (and cheapest) to impliment would be the 2 wheel / skid plates.
!*caution*! with the gyro, make sure its rated to a high rotation, when we tried to use that with our robot this year, it messed up because it could only do 80 degrees/sec !/*caution*/!
....

I think there are three vesions of the Analog Devices solid state gyro (yaw rate sensor) with different max turn rates, with the highest being 300°/S.

The faster sensor lets you spin your bot quicker with the loop closed, and it will stay 'locked' if your bot is hit, or hits something, and spins at a rate below 300°/S - but neither of these things are show stoppers.

the slowest sensor: 80° / S is a fast enough turn rate for normal driving. If you are spinning your robot 360°/ Second, you are not driving, you are doing a victory spin!

Two things: when the sensor is turned faster than its max rate, it outputs the full scale reading - your control loop will continue to respond with its max output, so the robot does not go berzerk, it acts predictably.

The only shortfall is if you are integrating the sensor to get compass heading, the reading will be wrong if the robot is spun too fast (by an external force). In this case you need a way to reset the heading (if the sensor is being used that way) during a match.

2. If you close the loop on steering with a gryo sensor, I recommend you have a disable switch on the control panel. That way you can drive with the loop closed, giving you very precise steering, and the robot will fight on its on to hold its heading (when something external tries to push it sideways), and then if you want to drive the robot open-loop (victory spins and stuff) you can, with the flip of the switch.

Ball casters solve this problem nicely (McMaster-Carr #2364T1).
...

yes, and ball castors or skid plates take up very little space, which means, you could also have pnuematic feet that extend down, with high friction plates, that can be deployed when you want your robot to stay where you parked it!.