Apr 25, 2007

4wd vs 2wd? A Drive System Comparison

A couple days ago, my FLL team was designing a robot for next year's robot game, when we came to an unanswerable question: Should we use 4wd or 2wd?
Being their mentor, I encouraged a more 'scientific' way of deciding by creating a chart, and this is what they came up with:

4 Wheel Drive
  1. Slippage occurs during turning
  2. Navigates more terrains smoothly
  3. Uses more battery power
  4. Capable of in place turning
  5. Goes straighter
  6. The robot has more support
  7. Must use gears / pulleys to transmit power but has unnecessary gear lash
  8. The chassis usually becomes a boxy shape

2 Wheel Drive

  1. Free-er turns
  2. It is more difficult to navigate all terrains
  3. Uses less battery power
  4. No in place turning
  5. More likely to drift
  6. Have to use a caster wheel or other means of support
  7. Could be directly connected to motor and not risk slop
  8. The chassis usually becomes a rectangular kind of shape

Both drive systems have been used successfully in the past. A good example are the two 1st Place Robot Peoformance Winners: 1221 Built on the Rock had a stable and versatile 4wd robot while 4 Mindstorms Troopers had a simple and speedy 2wd drive robot.

We would also like to ask you to contribute to our comparison. So comment away!

Robolab 2.9

9 comments:

Myles said...

What? No in place turning for 2 wheel drive robots? Ive built many of them and they all can turn within its own footprint.

Robolab 2.9 said...

I think what they meant that the pivot point for a 4wd is in the center of the robot wheras the pivot point of a 2wd would be between the two drive wheels

Robolab 2.9

Louise Dade said...

You can have the pivot point in the centre of the 'bot on 2wd vehicle too. Have two castor wheels: one at the the front and one at the rear.

Or even ball-castors at either end to reduce friction and slippage from the rotation of the castor.

There might still be performance and accuracy issues though. It might be worth an experiment?

Eric D. Burdo said...

Too bad tracks weren't faster. But I don't see them every being fast enough for timed competitions.

Brian Davis said...

I think some of these limitations/strengths might be based on unstated assumptions. for instance, "in place turning" can be done with both platforms - in fact, I find it's somewhat easier with two main wheels. Those two wheels generally are near the physical center of the robot due to where you want the center of mass anyway. And with a 4 wheel drive, slippage does not have to occur during turning (like at the MER for example - six wheels, not slip), and a 2 wheel drive can ge very straight (fixed caster, not turning, and both wheels tied together with a limited slip connection). I'm not sure I see why a 4 wheel drive would use more battery power either, and gear slop, while it could be an issue, might also be addressed using "split gears" as i've seen some folks do.

I suspect both drive systems have been succesful because the folks who use them understand the strengths and weakness of their choice, and use that information in planning the missions.

--
Brian Davis

Anonymous said...

Im using 6 wheel drive with 2 motors. Main reason is weight. Weight and caster wheels dont get along, in my opinion. I also may need to pull or push an object so I needed the extra traction. I am having good results but it is very "box shaped".

nicthegr

Dean Hystad said...

I think instead of 4WD vs 2WD you should use the terms Differential steering vs Differential Skid steering. I can build 2WD and 4WD robots with Ackerman steering, or have an articulated chassis like some big construction equipment uses. The number of driven wheels is not a good identifier.

Choosing between differential steering and skid steering depends on what is most important to you.

Used to be that teams picked skid steer so the robot drove straighter. It wouldn't turn very accurately, but there were ways to compensate (square up against a wall or don't do many, or any, turns).

With NXT it is pretty easy to go straight. So I would choose differential steering to make the turns accurate. Put most of the robot weight over the drive wheels and make sure the frame and drive train are structurally sound. Of course you still should use sensors and every navigation trick you've ever learned.

As for traction I don't think you get much benefit from using more than two wheels. I remember the "Volcano Panic" challenge very well. Robots had to climb up the side of a fairly steep cone. Tracked robots didn't stand a chance because they have lousy traction on a smooth surface. Skid steer robots did OK, but still had some slippage. The most successful were the differential steer robots that used the soft "dragster slicks" like tires.

It's kind of early to be working on next year's challenge though. We won't know until mid September what the requirements are. I am hoping for more varied terrain. This year's challenge was too easy.

To keep my team busy we are doing background on the research presentation and participating in a series of robot challenges. This month is Sumo. Next month is a maze runner. After that I am thinking either capture the flag or obstacle course.

Robolab 2.9 said...

I realize that the observations were really based on, as Brian said, unstated assumptions. But just remember that we are dealing with elementary school kids, and that they built 2 robots, each with the given drive train, and observed differences.

Louise,
You have a good point, but that usually takes up too much space for their liking.

Eric,
We actually won one year with tracks. One of the missions was to drop a ball into a basket before the opponents did. We went for a side shot, while most opponents went straight for it, and we still got it every time. We got the speed by using the fastest motors we had, and using the E2 Lithiums.

Brian,
I'll be sure to remember all those wise words. What's this MER? It sounds interesting. And the split gears?

Dean,
I completely agree with you, but again, remember these are elementary kids. I'll also be sure to remember your words as well.

I also remember Volcanic Panic, ha! We used a big two wheel drive to charge up the mountain, but because we had to buil ourselves, it was different than the one used in competition, and we missed it.

I too agree -- this year was too easy! I am assuming Scott made it unordinarily easy this year because of the amount of newbies that were expected because of the launch of the NXT and everything.

Thanks for all the comments! Keep them coming!

Robolab 2.9

Dean Hystad said...

MER is the "Mars Exploration Rover".

A split gear is a trick that allows you to preload a gear in both directions. It trades a bit of friction to eliminate slop.

The best way to envision a split gear is to take two normal gears and lay them on top of each other. to make a thick gear. Now connect the two gears using springs. Orient the springs so they make the gears want to rotate about the axle relative to each other. Looking at this gear along the axle it would look like a normal gear with two sets of teeth that were slightly offset.

When you mesh the split gear to another gear the teeth are forced into alignment. Aligning the teeth cause the spring to stretch or compress (depending upon design). This generates a force that tries to spread the teeth back apart. The result is that half the split gear will touch on one side of the gear teeth, and the other half will touch on on the other side.

Look at "Building LEGO Robots for FLL" to see an example of a split gear made out of LEGO. I twist an axle to make a torsional spring.

And as for the "remember, these are elementary kids" remark, I find I have the most success when I forget that my FLL team is comprised of fourth grade girls. If you expect little that is what you will get.

Related Posts Plugin for WordPress, Blogger...