Locating lost cattle on large ranch

Posted by Greg Kosin on October 9, 2014 at 8:13am

I happend to have lunch with a cattle rancher and somehow the topic of flying my Multicopter came up. The rancher immediately  perked up and asked if I could fly and find his cattle. They graze on hundreds of acres and then round them up to move them or send to marked but he currently has a dozen or so missing. So I started thinking about how using a plane they could search for the missing ones. The concerns I have would be.

  1. Could you even identify the cattle from the air in either a video or still pictures and how efficient would it be to have to either look at or watch a video to locate them. 
  2. What altitude do you think you could get decent pictures from to be able to identify cattle easily yet cover the most amount of ground at once?
  3. The gps coordinates would need to be overlaid onto the pictures or video to know where the picture was taken. 
  4. The cattle will go under tree cover during parts of the day so he said early morning or late in the day they would go out and graze in the open so most operations would need to be early in the day or late in the day. 
  5. How fast could you fly or how much area can be covered in say a half hour flight. ( I don't know how long a plane can fly)
  6. This land is in the foothills so terrain can be open flat and in the middle a hill etc so altitude awareness is a must. 
  7. If all of the above things can be sorted out, would this be something a ranch hand employee of his could learn to do reliably to make it worthwhile?

I know we have a bunch of really smart people here so give me your thoughts on this. I fly my Y6 for fun and have a gimbal and gopro but the vast area I have to believe requires a plane due to speed and flight times. Attached is a picture of what some of the terrain looks like. 

Thanks

Greg

Screenshot_2014-10-09-07-55-37.png

You need to be a member of diydrones to add comments!

Join diydrones

Email me when people reply –
Follow

Replies

  • Dennis Okeefe October 9, 2014 at 9:18am

    i have been working with a desert search and rescue group to develop a "system" to locate lost "things" in terrain like you show in your picture.  we started out thinking multi-rotor but quickly abandoned that after we realized the limitations on range and flight time - too much land to cover with that approach.  we switched very quickly to a fixed wing airframe;  we used a calypso EP glider since one of the team had one available.  we added an apm autopilot and a camera or two and did some experimentation.  what we found was the range and flight time was fine but the payload was low so obvously we needed a different airframe.  we switched to a penguin and that is what we are currently flying.  we have used many camera configurations (gopro H2, H3, mobius, 808 cam series, etc) and while all give adaquate coverage for what they are pointed at we have a couple of issues.  first, the geo-location capabilities of those cameras is a pain to accomplish.  they don't have GPS and tagging the video is hard if not impossible in realtime.  they also only see what they are pointed at - i will get to that in a minute.  we experimented with stills as well as video and find the video MUCH easier for people to find things on quickly than stills.  remember, we are developing a "system" which means planning before flight to analysis of the data after the flight with the goal of QUICK location of the lost "thing"...

    in may of this year i discovered a project on kickstarter for a 360 deg camera that i thought had some very interesting capablities that might work well for our project.  it's the 360cam by giroptic and is just about to ship to supporters and will ship to everyone early next year. ( http://www.360.tv/ )  it will provide us 360deg live hd video (wifi) with battery capabilities that would easily outlast the airborne platform it was being carried on.  it needed to be mounted completely below the airframe to allow for the 360deg coverage - to me, that meant it needed to be retractable for landing (or other) ground impact since it was going to be rather expensive ($500) and i didn't want to destroy it on landing.  

    to solve the retractability issue i developed a retract mechanism specifically for this camera but a mechanism that will also work with most any other small camera.  with a little work i could make it function with a stabilized mount like the tarot 2 but i haven't pursued that since i don't have time.  the retract device requires a hole in the bottom of the aircraft about 3"x5" - pictures are attached.  i have built a prototype which we are about ready to fly in our penguin.  i also adpated the mount to our multi-rotor and combined it with landing gear assembly - keep in mind the need to keep the camera lenses (3 of them) below any part of the airframe including the gear.   the multi rotor version has also been built and ground tested and we are about to test fly it.  pictures are also attached of the multi-rotor application of the retract.  here is a link to more pictures...

    https://drive.google.com/folderview?id=0BwZ6hVSoB7e8VXRaVHR1WlJXXzA...

    we are waiting for our camera and getting a anaconda (http://www.readymaderc.com/store/index.php?main_page=product_info&a...) to mount the retract mechanism and camera in.  

    i will report the results as soon as we test this configuration.

    dennis

    3701848239?profile=original

    BLKMNT360C0001.jpg

    BLKMNT360C0002.jpg

    REDO3Plate.jpg

    • Greg Kosin > Dennis Okeefe October 9, 2014 at 9:29am

      Wow Dennis, just the time you have in all your drawings is impressive. You have to love the concept of Kick Starter to help inventors get their great ideas made. I have supported a few myself. I look forward to your post with your test results. 

      Greg

  • Dave Schilling October 9, 2014 at 9:16am

    Hi Greg,

    I am sure you could do what your asking with just long range fpv system. I have a Skywalker 1800 and with a 5,000mah battery I can easily fly for 20-30 min depending on throttle management. Its all about how much you want to invest but all things you asked about are definitely possible. The arduplane would be excellent for gridding autonomously. You can live stream from GoPro and easily tell what you are seeing. Check out youtube videos of fpv flights. You can go really high-tec if you are planning to do this commercially but in the USA you have lots of hurdles. I would suggest do it for free and maybe get some free beef in return.  

    • Greg Kosin > Dave Schilling October 9, 2014 at 9:32am

      Dave, I have mounted a FPV camera on my 3dr Y6 but have yet tried flying it using fpv. In my initial setup my video link is not solid. I have a 5.8 system to try out. I love the free beef idea. Maybe cut a deal like 10%. So every 10 steer I find I get one of them LOL

      Greg

  • MikeInFrance October 9, 2014 at 8:36am

    I think you can check the anti poaching initiative and how they have come to select some interesting aircrafts, although I think some donations might have had some influence.

    Play also some consideration to IR cameras, as cattle should stand out with infra red radiation vs the background.

    Regarding Detection, I think that taking 2 simultaneous photo streams, one with a normal cam and one with an IR or NIR cam would give good results, especially if coupled with GPS coordinates.

    Regarding sizing, the size of the area to be covered plays it's role, but the amount of data and the time to react since flight play their role too. Having detected the cattle on the pics is one thing, collecting it is another, since in the meantime it may have covered some distance.

    Ideally, I would see 2 reconnaissance flights: 1 for detection/planning, and 1 for crosschecking it's presence where it's supposed to be just before collecting it.

    The 2nd flight could be done with a smaller frame since it is more of a cross check. I could imagine it being done with a quadcopter, while for the 1st flight, I think a plane or wing would be preferable.

    • Simon Howroyd > MikeInFrance October 9, 2014 at 8:39am

      Just to add to Mikes good comment;

      We found we could detect humans really easily without needing IR. It completely surprised us to be honest as that was our original plan!

      Thought I'd throw my findings in :-)

    • Greg Kosin > Simon Howroyd October 9, 2014 at 9:20am

      Thanks for your quick response. I looked at your link and found it very interesting. I didn't see any video or information re you detecting humans and how you did it without IR. The heat signature of a cow would be rather large I would imagine. 

      Greg

    • Simon Howroyd > Greg Kosin October 9, 2014 at 9:24am

      No problem, thanks!

      We did all the image processing offboard, post flight using Matlab. I wasn't the SME for the object detection, I can find out, but I believe it was using colour boundaries. A lot of that is inbuilt to Matlab anyway.

      The range of IR is very limited so seeing any sort of distance with a wide field of view without needing a ridiculously expensive camera we found to be pointless. We used a GoPro for the visuals and took pictures at 2Hz (I think).

  • Adam Eng. October 9, 2014 at 8:33am

    hi

    yes most definetly can be done

    1) i use still images
    2) 80m
    3) yes and can be done
    4) correct
    5) 2 - 3 hours we can fly comfortably
    6) yes we can use some sensors but ideally you would want to use human intelligence here for the autonomous flight and plot the way points.
    7) yes would be but doesnt come cheap

    this is my current solution http://www.spotifLy.com

    regards

    Spotifly | Spotifly
    Spotifly | Spotifly
  • Simon Howroyd October 9, 2014 at 8:31am

    Greg,

    This is pretty much a search and rescue type mission.  Some of my undergraduates at Loughborough University made a good start on this type of project this year, here is the report. Please bear in mind that we are in the research phase, so a long way from commercial realisation, which is of course not the main effort of a University.

    To answer some of your questions:

    1. Yes, see the Loughborough project. It isn't easy though.
    2. We were flying at 400ft, the maximum legally allowed in the UK at the time. The lower you are the easier for the computer software to recognise an object, but the harder it is for the aircraft's endurance.
    3. Many cameras these days have GPS so can geotag images (embed the GPS location inside the JPEG's metadata). Processing images is easier than video, so if you find something of interest you know where it is from the geotag.
    4. If they are in a forest you're stuffed. If is is sparse trees, such that the UAV can see under the canopy when approaching then this is fine.  We found that our software detected humans well before we were directly overhead.
    5. We cruised at 55mph which is reasonably typical. In terms of endurance then it's like saying how long will a car drive on a full tank of fuel...depends on the car, size of tank, driving style. I cannot answer this question.
    6. Terrain following is handled already in some autopilots, eg ArduPlane/Copter. It inspects Google data when setting waypoint altitude and allows you to select "above ground" or "above home" (to paraphrase).
    7. Anyone can learn anything if taught well. Self teaching is hard. You may need to look at the legalities of commercial UAV operation in your country.

    Hope this gets you going!

    Simon

This reply was deleted.