DPRG
DPRG List  



[DPRG] Re: Skid(s) or Castor?

Subject: [DPRG] Re: Skid(s) or Castor?
From: David P. Anderson dpa at io.isem.smu.edu
Date: Thu Oct 18 00:40:01 CDT 2001

Howdy

Interesting discussion.  

My own experience with two-wheel differential drive robot platforms
has been mostly with tail-wheels. I did have a few lego robots with
tail skids.  

In both cases the reason for having either is for support and
balance.  That is, the tail device is not a powered wheel or
or a steering device, just a passive point of support.

As such it shares the load with the main drive wheels.  Too much
weight on the drive wheels and the robot is not balanced, easily
tips up on it's nose.  Too much weight on the tail and the drive
wheels lose traction and slip, especially in turns and rotations.

Borenstein (U.of.Mich) suggests that some common odometry errors
arise from too much weight on the tail with it's attendant friction,
causing the drive wheels to slip when turning.

This is the bot's center of gravity, or balance point.  SR04 balances
about 1/2" behind the main axle, well forward of the tail.  The lego
bot in its current enlightenment also balances about 1/2" behind the
main axle.

For either the tail wheel or skid the trick is to have the same
amount of friction in all directions.  This suggests that some kind
of omni-ball might be ideal.  For a standard castor, the friction
required to roll forward or backward must be very nearly the same
as the friction required to rotate the castor around it's center.

Try the following experiement.  1) Place the robot on a flat surface
and push it forward from a stopped position.  Get a feel for how much
force it takes to get it to move. (This assumes no motors engaged, with
free-spinning drive wheels)

                  ========
             -----------------
             |               |
 push -->    |               |
          ====       0       |
             |               |
             |               |
             -----------------
                  ========
            

2) Now with the robot stationary push it sideways at the rear, forcing
it to rotate around the center of the axle.  Get a feel for how much
force this takes to get it to move.

            push
             |
             |
             V
                  ========
             -----------------
             |               |
             |               |
          ====       0       |
             |               |
             |               |
             -----------------
                  ========

            
For a castoring wheel, this will make the castor rotate 90 degrees (more
or less) before the wheel itself begins to turn.   The two forces need to
be very close to the same, whether pushing straight or sideways.

That's what suggested that some sort of omni-ball might be the optimal
solution.   A simple skid might also have the same friction in all
directions.  The main problem with a simple skid is that it is hard 
to get it to carry much weight without building up too much friction. 
It works on the little lego bots because they are so light.  Skids also
tend to get hung on things in the real world (i.e., my house), whether
they are light or not.

The new bot's I've been working on have 3.75" drive wheels, close
to your 4" spec.  I'm running them off a 12 cell pack of NiMH 1800
mAH AA cells, and the robots are aluminium and lexan with bits of
steel and brass.  Heavy enough to require a tail wheel.  I recently
added ball-bearings to the tail castor for the very purpose of lowering
the friction required for turning.  Prior to that it was flunking the test 
outlined above.  The current design looks like this:

ftp://www.geology.smu.edu/pub/users/dpa/robo/nbot/ntw18.jpg

and it seems to work really well.  Rev. 3.0  (why does everything
take 3 revs?!)

As an aside, I'm always looking for things I can make that will allow
me to play with and learn more about my lathe and milling machine.  This
is not necessarily what you're looking for in terms of inexpensive and
easy to produce!

ftp://www.geology.smu.edu/pub/users/dpa/robo/nbot/ntw_x2.jpg

shows a layout of the individual parts for the tail wheel, axle, castor,
bearings, and support.  The full platform looks like this:

ftp://www.geology.smu.edu/pub/users/dpa/robo/nbot/ntw19.jpg

Not sure how helpful any of this is.  Mainly that the robot's center
of gravity must be chosen carefully, and the third support point must
have the same, or very close to the same, friction in all directions
of movement.

regards,
dpa


> I've had a leisurely debate echoing in my head of late about the
> merits of a castor tail-dragger wheel versus a skid or skid-plate
> on the back of a small-to-middlin' 2-wheeled roller -- kind of
> like a fire-fighting size chariot with 4" wheels.
> 
>        |----- 10" ----|               |----- 10" ----|
>                _---_                          _---_
>        =======/=====\==               =======/=====\==
>        |      |  o  |        vs.      |      |  o  |
>        =======\=====/==               =======\=====/==
>       ()       `---'                 _/       `---'
>
> Anyone have skid/castor experience (pro or con) they'd care to share?
>
> Favorable traits I'm looking for are:
>
>        ease of manufacturing (will be making many)
>        robustness in typical hobby Robot environment (huh?!?)  ;-)
>        cool factor
>        cost of manufacturing
>        minimum introduction of odometry error
>        (???)
>
>Por su atencion, gracias...
>

More information about the DPRG mailing list