19-01-2009 09:34

GeoffP

Re: pic: steering methods

this shows several simple steering methods if you are going that route. left/right (tank) control will not work well this year since friction is so low. remember that sliding (dynamic) friction is roughly half of not-sliding / rolling (static) friction. also keep in mind that velicity is not a variable in the equation for the force of friction. 4 wheel independent drive, limiting torque to just below the break point between static and dynamic friction, and positive steering will be the keys this year.

19-01-2009 10:17

KRUNCH DUDE

Re: pic: steering methods

car drive and 4 weel drive are too hard to machene and takes to long to build

19-01-2009 10:25

Bongle

Re: pic: steering methods

Many videos have been posted of weighted robots towing trailers on regolith using skid steer without trouble. The coefficients of friction in the manual may not be entirely correct.

Another problem with many of those suggestions is that the require ground-level motors and linkages, which take up room that could be used for a ball picker-upper. I don't know about you, but I'm of the belief that human-only-loaded robots will be at a disadvantage this year, in contrast to 2006.

Cool ideas, though. I might bookmark this page for ideas in future years.

19-01-2009 12:36

jgraber

Re: pic: steering methods

Very nice drawings, thank you.
Can you add one for wide drive, with independent linkages for front and back, or some other way to allow strafing right left?

I think left-right independent drive is enough, 4wheel not needed, although independent wheel motors are easier to angle.

Most posts on traction control mention limiting wheel slip, not checking for torque. What kind of sensor would you use to measure torque? Some kind of chain tension sensor?

19-01-2009 13:02

JesseK

Re: pic: steering methods

For the last two drawings (linkage drives), I believe there is a more practical way to link them when considering an FRC drive train. In order to keep the wheel base at its maximum, regardless of which way the wheel modules are oriented, the motor must always stay on the inside and the wheel on the outside perimeter of the bot. The easiest way to achieve this is to link diagonal wheel modules together rather than the modules on a side.

Great drawings though!

19-01-2009 13:45

DonRotolo

Re: pic: steering methods

Nice drawings. We are investigating if car-type steering will have any advantages with our chassis this year.

Several threads (and my own tests) have shown that the 2:1 ratio of transverse to lateral friction is a myth, reality shows it to be about 6:5.

19-01-2009 13:46

GeoffP

Re: pic: steering methods

you can play with the linkage positions to make the inside radius turning wheels turn more than the outside wheels (so neither have lateral slip during turning because the radii have the same center).
A simple way to make the independent drive wheel is use a castor, bolt the gb to the bottom/swivel of the caster along with a steering arm, then connect the wheel directly to the gb output shaft. then just attach the linkages or chains to the steering arm. 4 wheel steering is ideal and only requires one drive design X4.
I'm not familiar with traction methods but I would think you could monitor the current and see a sudden drop when the transition from static to dynamic is reached, the power to the motor could then bump down to 30% to re-establish traction and then brought to 90% as its new target.

19-01-2009 14:20

GeoffP

Re: pic: steering methods

the 2:1 ration i am referring to is static vs dynamic friction coefficients. a simple test would be to fix 3 wheels leaning towards each other like a tripod. then incline the field material and measure the angle they start sliding at. while they are sliding, decrease the angle until they stop and measure that angle. the ratio of the sine of the angles should approximately match the ratio of the coefficient of frictions static vs dynamic. it will also be your increase in acceleration if you can prevent your wheels from sliding either laterally or transversely.
static friction is intermolecular bonding between 2 stationary surfaces (even if one is rolling). dynamic friction is trying make those bonds while the surfaces are sliding plus includes momentum. they should be very different.
that being said, the ratio is much closer with some materials and especially at low friction so your testing may be correct. thanks for posting that.

19-01-2009 14:35

GeoffP

Re: pic: steering methods

if skid steering is used (2 left wheels in-line and driven 1 speed, 2 right wheels in-line and driven another speed), and the side robot is up against the wall where wheels on one side are on carpet and the wheels on the other are on the slippery field, then will the robot be able to away from the wall?another method would be flywheel steering where the one side is pushed up or down to turn the bot with the forces from the torque produced from trying to tilt the axis. that may not get you away from the wall either though.
i think articulated steering is necessary.

19-01-2009 20:14

gblake

Re: pic: steering methods

All of this is nice for a machine that is rolling in a straight line (no changes in the orientation of its velocity vector, just changes in the magnitude); but since most of these discussions center around the notion of using "traction control" during turns...

Just as soon as one of these slippery wheels is rotated (about a vertical axis) to try to turn a bot, I assert that the wheel has to begin slipping sideways; and further I assert that it will continue to slip sideways until the dynamic (NOT static) friction forces between the wheel and the floor wipe out the bot's forward velocity.

If I am right, then no amount of traction control attempts made after a wheel is turned will put the wheels back into the static friction domain until the bot's forward (its pre-turn direction) motion is gone. Once that forward motion is gone and the wheels are no longer skidding sideways, traction control can then help move the bot in its new direction.

Alternatively, traction control could be used to slow a bot before a turn is attempted. Doing that would minimize the time spent slipping sideways after a turn is begun, but it wouldn't do much to help you run away from a pursuing bot if you are trying to do some evasive maneuvering.

Am I overlooking something? If traction control isn't very valuable, I would hate to see teams invest more into it than is appropriate.

Blake

20-01-2009 07:17

GeoffP

Re: pic: steering methods

Great analysis Blake. I agree with the scenario you describe and your bleak outlook on the difficulties we will have with turning. Your arguement also suggests that skid-steering would be even more futile(?). at least articulated steering gets the thrust vectors in the right direction. simple traction control such as just limiting the accel and decel of the motors may be helpful in a straight or in the middle of a turn. if we try to stay just under the static friction force then the bot may sporadically "grab" and respond better to our drive commands. i forsee a lot of jacknife steering though with 5 other bots and 120 balls on the floor.