DPRG List  

[DPRG] grounding motor cases

Subject: [DPRG] grounding motor cases
From: Ted Larson ted at larsonland.com
Date: Thu Jul 7 12:06:24 CDT 2005


We are using an el-cheap-o GPS unit.  A Garmin 18LVC OEM unit.  I bought it
at GPS city for $89.  It is a WAAS corrected unit which helps.  We take the
output of the GPS, and use IMU+odometry to generate our own additional
corrections to de-drift the output.   We do all our own path planning.  This
is how we are able to get it to consistently be 1 meter accurate, and not be
wandering all over the place.  Without any correction the robot tends to
drive a long-S-shaped path following the drift all over the course.  This is
why it took us almost of year of fooling with the robot, trying different
equations, and such, to get it to work properly.  We still don't have the
dead-reckoning end of it working properly, because the GPS is biasing the
equation too much.  That is next on the list....all it takes is time.  At
least the weather is good outside for getting this work done.

I am trying to work the schedule out so that one of us will make a day trip
up for the Robomagellan event.  So, we can do both events.  I think it will
all depend on what airline flights are cheap, and what the schedule looks
like, if a day trip will be possible.  With it working as well as it is, I
really want to compete it up in Seattle this year.  One other
possibility...Bob and I...as crazy as we are....have actually contemplated
building a 2nd robot that is a slightly improved one of the first, so we can
be in two places at once.  Odyssey is almost 1.5 years old at this point.
We have learned a lot of things we would do differently if we were to build
him a sibling.  Again...it all boils down to having the time to do it.  I
never have enough time to do all the things I want to do.

I think if you show up in Seattle....that would be even more incentive to
go.  Or if you don't get jBot working in time....at least come to San Jose
for Robonexus.  Either way, plan on a robot event for that weekend.



-----Original Message-----
From: David P. Anderson [mailto:dpa at io.isem.smu.edu] 
Sent: Wednesday, July 06, 2005 12:59 PM
To: ted at larsonland.com
Cc: dprglist at dprg.org
Subject: Re: [DPRG] grounding motor cases

Hi Ted,

Well that is very encouraging, that you have been able to get that level
of performance from the GPS.  We've been sort of creeping up on it from the
other end, inertial navigation, and I think now the robot could maybe run a
robomagellan-type contest on it's IMU/odometry alone.  I haven't done much
with the GPS signal yet but it seems so unreliable at this scale, I'm
to find out that it can be made to work!  Are you using DGPS?  WAAS?

Well heck, Ted.  if we get this thing working by the fall and travel
all the way to Seattle for Robothon, you mean you might not be there? 


Say "Hey" to Bob.


> David,
> We have used both the Devantech compass, and the PNI Vx2E
> The PNI does have soft, hard iron calibration, and is a very sensitive,
> accurate device.  The problem we ran into with it, was we needed better
> motor shielding, based upon where the compass is mounted on our robot.
> motors are right angle, and the compass sits right in-between the
> bubbles of each.  Use a gauss-meter to look at where you plan to mount it,
> especially with the robot driving around on the ground under load, so you
> can see what the ideal mounting will be.  The Devantech works good in that
> position, and we get +/- 5 degrees out of it.  The PNI is good to 1 degree
> or better, but hops all over the place as soon as we get the motors
> I have done testing with it mounted up on a pole above the robot which
> eliminates the field problems.  HOWEVER, you really want your compass as
> to the ground as possible.  These compasses are not tilt-compensated.  The
> larger the moment-arm, the more tilt from bumps in the road, and the more
> the compass is swinging around from tilt issues.  So, this is why we are
> currently using the Devantech, and it just barely works well enough.  We
> hope to go back to the PNI once we shield our motors better...more on this
> in a sec.
> The ultimate is to either own a tilt compensated compass ($$$).  Or here I
> am again, going on the cheap for more robot fun.....build your own Z-axis
> magnetometer and tilt-compensate the compass yourself.  I think it might
> possible to tilt-compensate a cheap compass compass using a Z-axis
> magnetometer, and a 2-axis accelerometer together with a Kalman filter,
> pretty easy.  I have the math for doing it using just the Z-axis iron, and
> it is a nasty piece of calcs.  More on this later...all it takes is time I
> don't have. 
> Make sure whatever enclosure you build for your compass has NO FERROUS
> materials in it.  Ours is enclosed in a PVC box held together will all
> fasteners, and aluminum stand offs.  One steel screw somewhere is enough
> mess things up.
> I have been looking into building Mumetal shields for our motors.  That
> would be the ideal setup for keeping the motors from messing with your
> compass.  Here is a company that sells Mumetal foil that might work.  I am
> not sure what thickness we will need to block the field strength.
> need to spend some time with a gauss meter and doing some calcs to figure
> out what would help or not.
> http://www.lessemf.com/mag-shld.html
> One more thing on the subject of Sonar units to look at.  The main reason
> went to a separate battery for the sonar was that the 400v trigger, causes
> all kinds of power supply spikes on your logic.  It even bleeds back over
> the trigger and echo lines.  We have actually thought about just putting a
> PIC in there that does nothing but read Sonar, and then optically
> the whole thing from the rest of the robot.  You might want to make sure
> that isn't something that is going on for you.
> GPS....It has taken us MONTHS to get our GPS working correctly.  Issues
> noise, and bad data, and math round-off errors, and more....I think I have
> spent as many hours getting our GPS to work on our robot as I did trying
> get Bender to balance the first time.   I finally have it working REALLY
> well.  If I can get 5 satellites on a position fix, I can drive to within
> feet of the cone on GPS alone.  We did a recent Magellan challenge at our
> club last week, there were 4 cones set out on a course, and we touched 3
> of 4.  The only reason we didn't get the first cone correctly was because
> forgot to take the lens cap off the camera (not the robots fault).  We
> need improvements on obstacle avoidance, and handling areas where GPS
> reception is poor using our inertial unit, however, we are working better
> than ever at this point.  The only bummer is, that Seattle is the same
> weekend as Robonexus, so I am not sure if we will be able to compete it in
> Seattle this year.
> So...the answer is....GPS is possible to work well, but getting it to work
> perfectly (as one would expect it SHOULD work), is up there in difficulty
> building a balancing robot.  I figure you of all people will understand
> comparison...but, I also know you are up for the challenge! :-)
> -Ted
> -----Original Message-----
> From: dprglist-bounces at dprg.org [mailto:dprglist-bounces at dprg.org] On
> Of David P. Anderson
> Sent: Wednesday, July 06, 2005 8:44 AM
> To: dprglist at dprg.org
> Subject: [DPRG] grounding motor cases
> Hi
> Ted wrote:
> > David,
> > 
> > Did you notice the problem before you added the .1uf caps to your motor
> > cases?  I usually avoid doing the motor case caps unless the chassis is
> made
> > out of a conductive material (i.e. not plastic).  I think it would be
> > interesting to look at your ground with the caps on, and with them off,
> and
> > also with your ground clips on, and off.
> I will try this.
> > Are you putting GPS on your robot?  Ground-plane noise issues will get
> even
> > more particularly interesting (sticky), once you have a GPS antenna in
> > mix too.  We ended up with a huge aluminum shield on the top of our
> to
> > get our GPS to work right.  Steel would have been even better, but that
> huge
> > chunk of ferrous material really messed with our compass.
> What sort of compass are you using?  Does it allow for hard/soft iron
> calibrations?
> > We eventually went to three battery backs.  One for Motors, one for
> > and one for Sonar, with the grounds tied together at one point only,
> > near the batteries.
> > 
> > -Ted
> I have a single tie point for all the grounds, but I'm, using a single 24
> volt
> battery pack, with a 5v switching regulator for the logic supply, and a
> separate 7v
> regulator for the sensors and IMU.  The GPS is a 3 volt device and at this
> point has it's own pair of AA cells, (which is sort of a pain because I
> always
> forget to turn it off.)  I don't think the noise problems are due to the
> power, as I get the same results when powering the sonar from a separate
> bench supply.
> Ted, are you getting any useful information from the GPS?  I ask because I
> don't really think that I am...
> cheers!
> dpa
> _______________________________________________
> DPRGlist mailing list
> DPRGlist at dprg.org
> http://list.dprg.org/mailman/listinfo/dprglist

More information about the DPRG mailing list