Outdoor robot navigation - GPS or robot lawn mower fence?

I've bought a WildThumper, will get it in a week or so.

Q1: For autonomous navigation on my lawn, would you recommend GPS (have a PAM-7Q GPS) or using an existing buried wire fence for my Robomow RL-1000?
GPS might be easier to get started with, but is it reliable enough to prevent the robot to drive out on the street or in a ditch? Using the buried wire somehow seems safer but more complicated to build.

Q2: Can I use PropScope to find the frequency in the wire? Once I've found it, what circuit should I build?

With both options I'd need to have obstacle avoidance sensors to avoid flower beds, our outdoor pool etc. Laser sensor perhaps combined with ultrasound and/or IR sensor comes to my mind, have them in my toolbox.

Also have several Project and Activity Boards, compass module, LSMDS1 9-axis IMU, Ping, IR leds etc. If this project ever flies, it would be close to 100 % Parallax-driven :-)

Q3: Don't have the Sharp IR sensor, would IR even be reliably outside?

Q4: As I'll travel to Santa Clara in a week, I'd be able to order some parts from Parallax and have them delivered to my hotel, anything I'd need for this project?


Appreciating guidance and recommendations

- Thomas Vikstrom (banjo)
Location: A hop from the longest bridge in...Finland

Comments

  • 28 Comments sorted by Date Added Votes
  • I have never tried to build an autonomous navigation lawn mower but I would look into a ColorPal only because it should be able to sense the color green.
    Larry

    If the grass is greener on the other side...it's time to water your lawn.
  • jmgjmg Posts: 10,042
    banjo wrote: »
    Q1: For autonomous navigation on my lawn, would you recommend GPS (have a PAM-7Q GPS) or using an existing buried wire fence for my Robomow RL-1000?
    GPS might be easier to get started with, but is it reliable enough to prevent the robot to drive out on the street or in a ditch? Using the buried wire somehow seems safer but more complicated to build.
    Why not design for both ?
    GPS is rather too vague for lawn-edge precision, but it could qualify zones for the buried wire, to cut false positives.

  • lardom wrote: »
    I have never tried to build an autonomous navigation lawn mower but I would look into a ColorPal only because it should be able to sense the color green.
    Hmm, that's an idea. Not sure how good a Colorpal work outside and inevitable some distance from the grass.
    jmg wrote: »
    banjo wrote: »
    Q1: For autonomous navigation on my lawn, would you recommend GPS (have a PAM-7Q GPS) or using an existing buried wire fence for my Robomow RL-1000?
    GPS might be easier to get started with, but is it reliable enough to prevent the robot to drive out on the street or in a ditch? Using the buried wire somehow seems safer but more complicated to build.
    Why not design for both ?
    GPS is rather too vague for lawn-edge precision, but it could qualify zones for the buried wire, to cut false positives.
    Yeah, had similar thoughts once. If only I'd know how to get started with the electronics for sensoring the wire. A metal detector circuit is probably tricky to calibrate as the wire sometimes is visible, sometimes centimeters buried. Remains then to sense the current in the wire somehow.

    What I'll start with now is to just hook up my GPS to a Activity Board with a LCD-screen, take it outside and see how consistent readings I get at different corners of the lawn.

    FYI: Not building another lawn mower robot, already have one 10 years old, what I have in my mind was to make a smaller robot that would be airing the lawn, e.g. with a couple of nails. How to build the airing mechanism is a later part, the navigating part needs to be solved first for the whole project to make sense. Still snow and ice on the lawn, have 8 months time :smile:
    - Thomas Vikstrom (banjo)
    Location: A hop from the longest bridge in...Finland
  • Definitely both. You'll want a magnetometer, too. I am working on some crude GPS navigation code and without accurate heading it's useless. The GPS alone can't give you as good a heading. I have that BNO055 and it's really convinient.
  • I've earlier ordered the LSM9DS1 9-axis IMU Module + I have a compass module from before.
    Related to navigation, as this robot would 'work' only on our lawn, I thought I could pre-program the map of the lawn with the GPS-coordinates of each corner (there are plenty of them due to the diamond shape of our lot). 'Connecting' these corners would then give the boundaries inside which the robot should stay. This is feasible only if the GPS is consistent enough which it might not be.
    - Thomas Vikstrom (banjo)
    Location: A hop from the longest bridge in...Finland
  • ercoerco Posts: 17,922
    Cool project, banjo! Wow, if you'll be in Santa Clara, then take a day and drive to Parallax in Rocklin, it's only 140 miles away! https://www.google.com/search?q=santa+clara+to+rocklin+ca&rlz=1C1CHZL_enUS704US704&oq=santa+clara+to+rocklin+ca&aqs=chrome..69i57.12714j0j7&sourceid=chrome&ie=UTF-8

    I'll bet Ken & Team Parallax would roll out the red carpet and you'd easily win the man-mile award, coming from Finland!
    "When you make a thing, a thing that is new, it is so complicated making it that it is bound to be ugly. But those that make it after you, they don’t have to worry about making it. And they can make it pretty, and so everybody can like it when others make it after you."

    - Pablo Picasso
  • banjo wrote: »
    Hmm, that's an idea. Not sure how good a Colorpal work outside and inevitable some distance from the grass.

    I'd be surprised if the Colorpal worked well in this application. I think it's intended to find the color of things up close.

  • Duane Degn wrote: »
    banjo wrote: »
    Hmm, that's an idea. Not sure how good a Colorpal work outside and inevitable some distance from the grass.

    I'd be surprised if the Colorpal worked well in this application. I think it's intended to find the color of things up close.

    A low res camera would probably be a better choice but the interfacing might be a problem. What the world really needs is a camera chip that transfers the image to a block of memory that can then be accessed via SPI.
    In science there is no authority. There is only experiment.
    Life is unpredictable. Eat dessert first.
  • erco wrote: »
    Cool project, banjo! Wow, if you'll be in Santa Clara, then take a day and drive to Parallax in Rocklin, it's only 140 miles away! https://www.google.com/search?q=santa+clara+to+rocklin+ca&rlz=1C1CHZL_enUS704US704&oq=santa+clara+to+rocklin+ca&aqs=chrome..69i57.12714j0j7&sourceid=chrome&ie=UTF-8

    I'll bet Ken & Team Parallax would roll out the red carpet and you'd easily win the man-mile award, coming from Finland!
    Yeah, has been in my thoughts, my dear employer though wants me to work when I'm there (I don't have a clue why :smile:), so a bit tricky to fit in the schedule. My day job has nothing do with electronics etc. so also tricky with the 'business case'
    Duane Degn wrote: »
    banjo wrote: »
    Hmm, that's an idea. Not sure how good a Colorpal work outside and inevitable some distance from the grass.

    I'd be surprised if the Colorpal worked well in this application. I think it's intended to find the color of things up close.
    That's what I understood also. However, if I put in another order to Parallax, I'll buy a Colorpal, can perhaps utilize it for navigation indoors if not outdoors.

    - Thomas Vikstrom (banjo)
    Location: A hop from the longest bridge in...Finland
  • ercoerco Posts: 17,922
    kwinn wrote: »
    A low res camera would probably be a better choice but the interfacing might be a problem. What the world really needs is a camera chip that transfers the image to a block of memory that can then be accessed via SPI.

    It might also be worth taking a look at Phil's Redeye project, using a linear imaging sensor: The Parallax/TAOS TSL1401-DB is a vision sensor that sees in one dimension using a 128x1 array of light sensors, along with an imaging lens.

    http://forums.parallax.com/discussion/117941/redeye-a-linescan-imager-based-boe-bot-line-follower

    "When you make a thing, a thing that is new, it is so complicated making it that it is bound to be ugly. But those that make it after you, they don’t have to worry about making it. And they can make it pretty, and so everybody can like it when others make it after you."

    - Pablo Picasso
  • erco wrote: »
    kwinn wrote: »
    A low res camera would probably be a better choice but the interfacing might be a problem. What the world really needs is a camera chip that transfers the image to a block of memory that can then be accessed via SPI.

    It might also be worth taking a look at Phil's Redeye project, using a linear imaging sensor: The Parallax/TAOS TSL1401-DB is a vision sensor that sees in one dimension using a 128x1 array of light sensors, along with an imaging lens.

    http://forums.parallax.com/discussion/117941/redeye-a-linescan-imager-based-boe-bot-line-follower
    Thx, cool project!
    I'm not however sure how I could utilize the linear imaging sensor in my outdoor project. The buried wire is...well buried and if I put another wire on the surface it will either get buried within some months or cut into pieces by the robot lawn mower. OTH, if I find some type of wide, somewhat flexible and thin enough material with suitable color and that would not easily get buried, this type of sensor could be useful. Don't know how aesthetic it would be though...

    - Thomas Vikstrom (banjo)
    Location: A hop from the longest bridge in...Finland
  • If the wire is not buried too deep you may be able to apply an AC signal to it and detect it when you are over the wire. That's generally how the electronic pet fences work.
    In science there is no authority. There is only experiment.
    Life is unpredictable. Eat dessert first.
  • ercoerco Posts: 17,922
    I suggests the linear imaging sensor as a colorpal alternative. I thought you were considering optically following the edge of the grass along a sidewalk.
    "When you make a thing, a thing that is new, it is so complicated making it that it is bound to be ugly. But those that make it after you, they don’t have to worry about making it. And they can make it pretty, and so everybody can like it when others make it after you."

    - Pablo Picasso
  • A ColorPAL -- or several of them -- would probably be the better choice. The TSL1401 is a grayscale sensor. Even with a color filter, it might not be able to distinguish grass from concrete. The ColorPALs do need to be in close proximity to what they're sensing, though. I would mount them on a hinged "sled" of some sort, so they can just skim over the grass tops as the mower moves along.

    -Phil
    “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away. -Antoine de Saint-Exupery
  • banjobanjo Posts: 236
    edited March 30 Vote Up0Vote Down
    erco wrote: »
    I suggests the linear imaging sensor as a colorpal alternative. I thought you were considering optically following the edge of the grass along a sidewalk.
    Ok, I see what you mean. There are gravel and asphalt sidewalks in some places, some areas are just bordering a forest or hedge between our and neighboring houses. So there's some contrast between grass and other surfaces in some places, but not in all.
    kwinn wrote: »
    If the wire is not buried too deep you may be able to apply an AC signal to it and detect it when you are over the wire. That's generally how the electronic pet fences work.
    Actually the wire is being used by the robot lawn mower so don't want to touch it, not sure what type of signal is used (brand Robomow RL-1000). Can I perhaps use a Propscope to detect the signal? Or just a multimeter if it happens to be a constant DC?
    To detect the wire signal from the robot, I'd need to build some type of circuit, perhaps including a 'antenna coil'?
    - Thomas Vikstrom (banjo)
    Location: A hop from the longest bridge in...Finland
  • banjo wrote: »
    ...........
    Actually the wire is being used by the robot lawn mower so don't want to touch it, not sure what type of signal is used (brand Robomow RL-1000). Can I perhaps use a Propscope to detect the signal? Or just a multimeter if it happens to be a constant DC?
    To detect the wire signal from the robot, I'd need to build some type of circuit, perhaps including a 'antenna coil'?

    It is most likely an AC signal so you may be able to pick it up with an AM/FM radio or a Propscope with a simple antenna. Might even have the frequency listed in the manual.
    In science there is no authority. There is only experiment.
    Life is unpredictable. Eat dessert first.
  • kwinnkwinn Posts: 7,479
    edited March 30 Vote Up0Vote Down
    PS - see here for a description of the signal.

    Enough information there to build a perimeter sensor for your project I think.
    In science there is no authority. There is only experiment.
    Life is unpredictable. Eat dessert first.
  • kwinn wrote: »
    PS - see here for a description of the signal.

    Enough information there to build a perimeter sensor for your project I think.
    Interesting read, explains pretty well the perimeter switch and the signal. Does not explain the sensor circuit in the mower as far as I can see.
    Found a patent that might not be for this exact model but at least from same company (Friendlyrobotics). A schema on pg 23 seems to cover the signal sensor which by the way also can measure the strength of the signal even far away from the wire.
    Now I'm mainly interested in knowing if the robot passes over the signal wire or not, the signal strength is of less importance.
    - Thomas Vikstrom (banjo)
    Location: A hop from the longest bridge in...Finland
  • jmgjmg Posts: 10,042
    banjo wrote: »
    ...
    Now I'm mainly interested in knowing if the robot passes over the signal wire or not, the signal strength is of less importance.
    Yes and no - I'd say you need to work 'on one side of the peak', in order to know you remain inside the wire.
    There also needs to be some tolerance for other items like metal pipes or grates, or simply cable depth.

    kwinn wrote: »
    PS - see here for a description of the signal.

    Enough information there to build a perimeter sensor for your project I think.

    Yes, good to see they used a crystal, which makes the chirp very precise and suits a narrow band analog filter (that can dig into the noise floor).
    Probably easiest to make one of those with another crystal, and a switched-capacitor filter/detector, tied to a sensor head that is broadly resonant at the 8kHz.

  • jmg wrote:
    Probably easiest to make one of those with another crystal, and a switched-capacitor filter/detector, tied to a sensor head that is broadly resonant at the 8kHz.

    You really don't need an external filter. The Propeller can do the filtering all by itself:

    http://forums.parallax.com/discussion/133173/fir2pasm-automatic-fir-filter-code-generator/p1

    And a narrow bandpass filter for 8 kHz should be a piece of cake.

    -Phil
    “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away. -Antoine de Saint-Exupery
  • Lots of advantages to using the Propeller for this. With the proper antenna arrangement you could calculate the position and angle of the robot in relation to the signal wire.
    In science there is no authority. There is only experiment.
    Life is unpredictable. Eat dessert first.
  • jmgjmg Posts: 10,042
    jmg wrote:
    Probably easiest to make one of those with another crystal, and a switched-capacitor filter/detector, tied to a sensor head that is broadly resonant at the 8kHz.

    You really don't need an external filter. The Propeller can do the filtering all by itself:

    http://forums.parallax.com/discussion/133173/fir2pasm-automatic-fir-filter-code-generator/p1

    And a narrow bandpass filter for 8 kHz should be a piece of cake.

    Well, maybe on paper...
    The issues arise when you seek to tunnel into the noise.
    Running an oversampled ADC and then hoping to find a signal below the noise, runs into LSB quanta effects.
    If you instead use external analog elements, that quanta effect is not present until you apply the ADC.
    The type of filter/sync detector I have in mind, gives I & Q signals that are envelope based, but can be built with a Prop's timers and simple analog switches (4066 class). The ADC applies to the post-filter/envelope detector.

    Nett result: Better noise floor.


  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 20,974
    edited March 31 Vote Up0Vote Down
    In that case, an I/Q detector can easily be programmed in the Prop without external components:

    http://forums.parallax.com/discussion/105674/hook-an-antenna-to-your-propeller-and-listen-to-the-radio-new-shortwave-prog/p1

    8 kHz is slow enough -- and the integration times long enough -- that the S/N should be quite high.

    -Phil
    “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away. -Antoine de Saint-Exupery
  • jmgjmg Posts: 10,042
    In that case, an I/Q detector can easily be programmed in the Prop without external components:

    http://forums.parallax.com/discussion/105674/hook-an-antenna-to-your-propeller-and-listen-to-the-radio-new-shortwave-prog/p1

    8 kHz is slow enough -- and the integration times long enough -- that the S/N should be quite high.

    -Phil

    Sounds worth trying, and it may prove 'good enough'.

  • Banjo,
    I have also done some thinking on small scale outdoor localization similar to your lawnmower robot application.
    GPS is too rough, but Differential GPS can give sub-100mm accuracy. There is a nice hackaday article exactly this issue;
    http://hackaday.com/2017/03/01/where-are-the-autonomous-lawnmowers/
    Alternatively, you can try some sort of triangulation beacons similar to the Vive Lighthouse interior system (the problem would be sunlight interference);
    http://www.theverge.com/circuitbreaker/2016/7/6/12109880/htc-vive-lighthouse-hack-diy-positional-tracking
  • ercoerco Posts: 17,922
    I'm surprised Ken hasn't chimed in. He's knee deep in GPS and Blockly for the Propeller, sharing regular updates in Facebook.
    "When you make a thing, a thing that is new, it is so complicated making it that it is bound to be ugly. But those that make it after you, they don’t have to worry about making it. And they can make it pretty, and so everybody can like it when others make it after you."

    - Pablo Picasso
  • Saw Parallax have a new GPS for sale. This one is much more affordable, but I already have a PAM-7Q GPS.
    macrobeak wrote: »
    Banjo,
    I have also done some thinking on small scale outdoor localization similar to your lawnmower robot application.
    GPS is too rough, but Differential GPS can give sub-100mm accuracy. There is a nice hackaday article exactly this issue;
    http://hackaday.com/2017/03/01/where-are-the-autonomous-lawnmowers/
    Alternatively, you can try some sort of triangulation beacons similar to the Vive Lighthouse interior system (the problem would be sunlight interference);
    http://www.theverge.com/circuitbreaker/2016/7/6/12109880/htc-vive-lighthouse-hack-diy-positional-tracking
    This is very interesting, can't though spend all my money on this hobby. If you try it out, please let us know
    - Thomas Vikstrom (banjo)
    Location: A hop from the longest bridge in...Finland
  • jmgjmg Posts: 10,042
    banjo wrote: »
    Saw Parallax have a new GPS for sale. This one is much more affordable, but I already have a PAM-7Q GPS.
    Cool, it seems to have 1pps, tho the vendors wiki is garbled.
    Has anyone compared this 1pps, with the 1pps from say Adafruit or other GPS ?
Sign In or Register to comment.