Drive Trains

[BigJ]

From a software standpoint, if you are deciding to go with a holonomic system (i.e., mecanum or all-omni), your programmers better be ready to learn some physics .

Team 1675 used a mecanum drive for the second time this year, and the programmers have a blast with it. I had so much fun last year just figuring out the way the vectors could change the direction. This year, we worked with the students to develop an algorithm based off an article about all-omni systems in SERVO magazine.

Also, have the programmers work with the drivers to make the controls more comfortable. The programmers can make the controls more or less sensitive (depending on how they are already).

[AdamHeard]

I'd say the drive performance you see on the field is 75% driver and 25% the details and craftmanship of the drive.

I've seen 6WD perform amazingly, and terribly.
I've seen Mecanums/omnis do the same.
Same for crab, same for tank tracks and so on.

End all, I think a well built 6wd is what lower resource teams should shoot for as the standard; It performs great and is the most applicable to most games. When 2nd picks come around, often times teams just desperately need a 3rd robot with a decent drive, and most often there isn't one available.

But, no matter what, if your driver is driving the current robot for the first time at the regional, you won't be preforming great. I'd highly recommend a practice robot (or at least a base) for driver practice at home.

[Madison]

Almost none of the specific details of a drive train are relevant to its performance on the field. In 99% of circumstances, wheel type, speed, or torque are inconsequential.

• A successful drive train is reliable -- it functions well all the time. If it breaks or malfunctions, it can be repaired quickly and easily.
  
• It is efficient -- it is well lubricated, quiet, and makes good use of its weight and power. It turns.
  
• It is no more or less complex than it needs to be -- it needs to work for your gameplay strategy. Adding bells and whistles that are beyond the scope of your strategy is a sure-fire way to shoot yourself in the foot.

That's it! It's more easily said than done, of course, but very little else about drives matters if you haven't yet mastered these three things. Once you've got this stuff down cold, you can start to think a bit more about the dynamic operation of the system and what changes and improvements can be made to make the most of that remaining 1% of circumstances when power, speed and torque matter.

[Rick TYler]

KISMIL: Keep it simple, make it last.

[Shadow503]

I loved our drive train this year. It was simple and always the fastest on the field! We have a rear wheel drive powered by 2 cims in direct drive for each wheel. We have powered ackerman steering in the front and software differential in the back. It is innovative yet simplistic XD

I have yet to see a robot out-drive us.

[dtengineering]

Originally Posted by BigJ

Also, have the programmers work with the drivers to make the controls more comfortable. The programmers can make the controls more or less sensitive (depending on how they are already).

This is an important comment that should not be lost amongst all the other valuable tips on the mechanical aspects of the drivetrain. The human-robot interface can make a good driver great.

Although we have used the "default" KOP joysticks and mixing code in the past, this year we knew we would have to have a much higher-performance drive system than in previous years.

We put a gyro on the robot and used it to make sure that the robot went straight when it was told to. Then our programmer re-mapped the turning axis to provide exponential response (very little near the "centre" but full-bore near the extreme edges). Finally we invested in a USB dongle from IFI, and hooked up a logitech game controller. Now the driver controls forwards and backwards with one thumb and left-right with the other. This is an advance over a single stick, as there is no chance of telling the robot to turn when you want it to go straight.

Most of this was accomplished using an older robot in the first two weeks of build, while the new drive train was being built.

Strangely, we found our new drive train performed well, and were really pleased with our driving.... even though mechanically we have almost the simplest drive train possible (modified KOP frame, four CIMS, two toughboxes, driving two IFI traction wheels at the front with two AM omnis free-wheeling at the back.)

People always tend to think of the programmers when they see a great auto/hybrid mode... but behind every great driver is some great drive code.

Jason

[MrForbes]

Good point Jason, this year we were playing with Ackermann drive and so we connected an R/C car control to it (the type with a small 2" diameter steering wheel and trigger throttle, all in a pistol grip) and it worked very well. Then we had an "aha!" moment and asked the programmers to make the new controller work with last year's 6wd drivetrain...and the result is an intuitive way to drive a skid steer robot. And a Rockwell Automation Innovation in Control award!

[BornaE]

Originally Posted by squirrel

Good point Jason, this year we were playing with Ackermann drive and so we connected an R/C car control to it (the type with a small 2" diameter steering wheel and trigger throttle, all in a pistol grip) and it worked very well. Then we had an "aha!" moment and asked the programmers to make the new controller work with last year's 6wd drivetrain...and the result is an intuitive way to drive a skid steer robot. And a Rockwell Automation Innovation in Control award!

That is what we did to test out our driving system with our 2007 bot to simulate ackermann drive that is on our 08 bot for testing hybrid. the code was also used on our back wheels as the electronic differential.