[TriEmbed] Interesting Kickstarter open source indoor/outdoor xyz ranging modules

Tadd Torborg tadd at mac.com
Thu Jul 28 11:24:59 CDT 2016


I wonder how hard it would be to use audio from the lawn-mower to the anchors to do time-of-flight distance measurement?  You could synchronize the lawn-mower and anchors with an ISM band message.  That could be used to wake up the microphone and audio receiver on the anchors.  Now a really loud two tone packet message, start at one frequency, then shift to another frequency.  That way the edge could be found in only half a wave cycle.  Gotta choose both to not be in the same audio bands as the lawn mower. 


<humor>I'm thinking a 100watts of 440hz switching to 880hz once per second would be interesting.  Do that faster as the mobile station gets close to a border.  </humor>



I'm sure there is a frequency which wouldn't kill the humans or dogs outright but would enable a decent location solution.  Sound is MUCH easier to deal with than light-speed range-finding.   And it's not like putting out a periodic pulse of noise on a lawn mower is going to render complaints.  It's not like silent running to start with.   It would be good if you could have a nice audio filter to get the extraneous noises out of the receiver.  At 10khz the sample rate of your AtoD could be practically glacial.  Heck, for prototyping we could use DTMF (touch-tone).  Just think of how much NOISE you could make haha.

Tadd


On Jul 28, 2016, at 12:09 PM, Rodney Radford <ncgadgetry at gmail.com> wrote:


Ken, 


A good first start is understanding the cause of GPS error and this Wikipedia article is a good summary (it is flagged as possibly in error, but to me, it is a nice easy to read summary of what I have seen in other places).


https://en.wikipedia.org/wiki/Error_analysis_for_the_Global_Positioning_System



So from that article, under a full sky with no bad weather, we can get about +/- 5m accuracy. Add in bad weather, trees, buildings, power lines, and the inaccuracies climb.


Sitting still and averaging the GPS readings will make this better, but whether it is enough is dependent upon what you need. If you are looking to determine if you are going to miss a tree or a building or property line, yes, you may be able to get to that level of accuracy.  If you are trying to cut a straight line parallel to another line cut earlier, I doubt it.


Ask any geocacher who has had to deal with getting exact GPS readings and they will tell you of the fun of sitting still and trying to average out the values. One of my worst geocaching memories was one of my first as I was trying to find a geocache before a storm rolled in.  While it was not raining yet, the clouds were dark, and my Garmin 60csx (one of the best consumer GPSs at the time) was showing +/- 150' swings as I stood still.  I finally gave up and returned another day.  But even on a good day, i would sometimes stand still for several minutes getting several hundred readings before the accuracy (as reported by the GPS) drops into the 2-3m range.


Many of the errors in the article above can be resolved with two GPSs, each in communication with each other, passing readings (and the timestamp). This is what differential GPS is all about.  


https://en.wikipedia.org/wiki/Differential_GPS

http://hackaday.com/2013/08/05/centimeter-level-precision-gps-for-500/



A good commercial differential GPS system can get you down to 10cm of accuracy - definitely enough to control a lawnmower, but they are not cheap.


If you google for DIY differential GPS solutions you will find a lot of links, but most are asking is it possible, or speculating ways it should be possible, but with very little real answers.


I played with creating my own DGPS system about 10 years ago using some $20 USB GPS receivers being dumped by Radio Shack.  I was using a Wifi system between a laptop and my home base and finally gave up as I was not making sufficient progress (and I lost interest).  But now with GPS receivers even smaller and with the advent of systems like the Raspberry Pi, this all can be miniaturized and may be worth dusting off the old code and trying again.


The layering of that with the Kalman filter approach would make a great system.  Use the DGPS data to get you close (assuming you would not get 10cm from a hobby implementation) and add in the data from a 9DOF accelerometer, magnetometer and gyroscope and you should get very accurate readings on position, direction vector and velocity.




















On Wed, Jul 27, 2016 at 9:32 PM, Ken Boone via TriEmbed <triembed at triembed.org> wrote:


I already have a lawn mowing robot test bed I built back in the 80s.  The optical beacon navigation i designed proved to be too inaccurate. The mower is described on kensrobots.com.

After a while when bench testing or displaying my fixed wing UAV the ground station displays a wobbly path taken around its fixed location.

A question?    For mowing navigation could a fixed GPS receiver at my house be used to increase mower location accuracy with an identical receiver on the mower.

First average fixed location readings to establish the fixed receiver location.  Then use differences to correct mower location?

I've been experimenting with the mower because I have grass and it is a neat challenge.  Too much liability envolved in mowing to want to develop  a product.
Ken

On Jul 27, 2016 5:17 PM, "Tadd Torborg" <tadd at mac.com> wrote:
>
> Firmware has to be pretty stable and well done since the transponder end has to turn around and answer.   Counter/timers needs to be very good/fast/well-understood.
>
> There are other off-the shelf solutions already.  Look for Decawave TREK1000.    I don't think Decawave is open source. 
>
> Range will be like cheap Wifi stations talking to each other, i.e. no nice antennas.  Good enough for lawn mower if you have two anchors line of site visible to the lawn mower at all times.  Put the intelligence on the lawn mower or in the house, and make solar powered anchors, look like driveway-light pedestals. 
>
> Tricky project.  I'm less scared of the range finding than I am of the robotics part of running a lawn mower around, so maybe it's not so hard.  Anybody want to get together on the lawn mower?  
>
>    Tadd


>
> On Jul 27, 2016, at 09:32 AM, Ken Boone via TriEmbed <triembed at triembed.org> wrote:
>
>> Kickstart info below this comment.
>>
>> I started out experimenting/learning  with a Knight-Kit 10 in 1 vacuum tube kit.
>>
>> It's amazing what technologies and resources are now available for students, hobbyists and garage/basement experimenters/inventors.
>>
>> More on the ranging modules.
>>
>> The modules uses fixed location nodes for xyz ranging of moving nodes or for your swarm distances between moving nodes.
>>
>> Wonder what's the maximum range? 
>>
>> Might need to resurrect my robotic lawnmower which needed a better ranging system.
>>
>> From site: https://www.kickstarter.com/projects/subpos/subpos-ranger-indoor-positioning-system/description
>>
>> "The Ranger Boards accurately determine the distance between one and other through phase shift/difference measurement and accurate timing techniques in the onboard Atmel AT86RF233 chipset. This is similar to how laser range finders work, except instead of light it uses 2.4GHz radio frequencies. The multiple antennas on board allow for multiple out of phase measurements to mitigate multipath effects, to best determine the distance between the Ranger Boards."
>>
>> Create - dream, imagine, experiment - Create. .............
>>
>> Ken Boone
>>

>> _______________________________________________
>> Triangle, NC Embedded Computing mailing list
>> TriEmbed at triembed.org
>> http://mail.triembed.org/mailman/listinfo/triembed_triembed.org
>> TriEmbed web site: http://TriEmbed.org






_______________________________________________
Triangle, NC Embedded Computing mailing list
TriEmbed at triembed.org
http://mail.triembed.org/mailman/listinfo/triembed_triembed.org
TriEmbed web site: http://TriEmbed.org



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.triembed.org/pipermail/triembed_triembed.org/attachments/20160728/2817452f/attachment.htm>


More information about the TriEmbed mailing list