Announcement

Collapse
No announcement yet.

Beach Detecting Rover project

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Beach Detecting Rover project

    Hi all,

    First post here. Awesome forum btw, really good posts all around! I've found a new favorite spot possibly..

    I've long had an idea for a project, that I have been kicking around. The project is to design and build an tele-operated semi-autonomous detecting rover, which will drive up and down a beach, all the while scanning the ground it traverses for metal, and uploading finds to the operator's remote control PC laptop.

    I had been looking for a while for a good metal detector design that I could incorporate into the design, and have found several good ones on this site. The idea is to have the detected locations wirelessly sent to a laptop with a nice GUI front-end, that is using Google Maps to display GPS coordinates. To help with the project, I am looking for any metal detecting source code that you could provide, to help me to in designing the software front end for the detecting portion. All of the software-related metal detectors I have found on the site are only offering hex binaries for the particular chip the designer chose for his detector, which won't work for me, unless I reverse engineer the binary (not gonna happen btw )

    Anyways, has anyone on the forum done any similar project? From the detector side, I'm debating whether the detector will be motion or non-motion. I can easily use motion, but it will require an axis with an oscillating servo, that will allow me to precisely control a good scanning range and motion.

    I plan on offering this design open source on an as going basis, to allow for others to contribute, and make suggestions. I plan on not only offering the entire design for the detector, but for the rover, and all hardware and software involved from the PC to the detector itself. This includes Eagle PCBs and schematics, CAD drawings, source code, etc. Nothing will be hidden.

    I know it's a very ambitious project, but with my background I should be able to get pretty far. I will even design printed circuit boards for any custom boards that need to be designed. Once the design is finished, I could even offer the boards and parts needed in a kit form of some type.

    As a quick rundown on the hardware and software I plan on using for the build, this is what I'm looking at using so far:

    1) Parallax Prop for geolocation sensor front end interfacing. Will try to use 12blocks if possible for non-programmers.
    2) XMOS DSP metal detecting front-end (or Prop II when available) Hopefully can build on XPI project.
    3) Flowstone DSP for PC-Based GUI frontend. Some pretty good high end DSP audio processing available here, now with Ruby language option.
    4) Gas Powered power source (small 1kw generator). Who needs batteries?
    5) Full sensor suite, i.e GPS, accelerometer, gyroscope, digital compass, PING ultrasonic sensors, etc.
    6) Wireless interfacing from PC laptop to Rover
    7) Phidget hardware for motor drivers, on-board I/O control, etc. SBC2 running Debian Linux.
    8.) Xbox 360 wireless controller operator frontend. Rumble control for detected metal.
    9) HTPC media PC running Windows XP or 7 for Flowstone
    10) Mbed LPC1768 or LPC 11U24 for USB virtualization hardware.
    11) Android based Cell phone for 3G internet connectivity. Could provide sensor data from here is well.
    12) 80/20 Extruded aluminum frame for hardware mounting
    12) Anything else I can throw in on this which would help..

    I've already assembled a large part of the hardware for the build. Today, I ordered the Barracuda kit today from Silverdog for testing the metal detector electronics.

    Individually, I've mocked up almost all of the pieces I need to build it. My biggest unknown right now is the metal detecting portion. I've been metal detecting on and off for twenty years, but I have not seen a lot of transparency in the metal detecting hobby, that you typically have in other areas such as hobby microcontrollers and such (Arduino, Prop, Xmos, MBed, etc). I have included links to some of the hardware I've discussed.

    Here are the links for the hardware:
    http://mbed.org/
    http://www.parallax.com/tabid/407/Default.aspx
    http://12blocks.com/
    http://www.xmos.com/
    http://www.dsprobotics.com/index.html
    http://www.parallax.com/tabid/597/Default.aspx *power plant only*
    http://www.zotacusa.com/zbox-nano-id61-plus.html
    http://www.phidgets.com/
    http://www.miymd.com/index.php/projects/xpi/ <-- hopefully..
    http://www.8020.net/

    Feel free to let me know what you think. I will update this thread as the project moves forward. I'm open to any and all suggestions.

  • #2
    So you're are proposing an automated target mapper?
    How about this... Mark ourt a 50 foot by 50 foot area in the sand and manually map out several quarters that you place in the square. Now go over it with the detector in the means you think your rover will go. See how accurate you are.
    Now give someone (your kid) the map and have them get the quarters back for you.
    Last step, take the detector and try and find the quarters again and dig them up.
    The success rate of each step will suggest the practicality of the concept.
    I suspect mapping is too inaccurate.
    I think we all have thought of driving down the beach in a Rhino and marking target spots with an automated detection gizmo. But then we realize the digging of the target is the hard part.
    I think if you are going to go to all this trouble with the electronics you should do an underwater detection system for gold in the Bering Sea. Should sell well.

    Comment


    • #3
      Interesting project.

      I have thought about starting a similar project for prospecting remote desert areas.

      Unfortunately, the difficult terrain and lack of 3G cellphone coverage, stretched the idea too thin.

      Will you be using a PC platform for development work? Linux /windows? Which uC will you be using?

      Comment


      • #4
        Here's a project a friend of mine developed a few years back. It maps out targets on a palm computer. I did a short test, and concluded that GPS is not accurate enough for coin sized objects. At least civilian GPS. It was like a +/- sq meter.

        If you had a 1 meter coil and were looking for larger targets it might be very useful.

        http://www.magsurvey.com/MetLSurveyHome.htm

        Regards

        Comment


        • #5
          Originally posted by Altra View Post
          Here's a project a friend of mine developed a few years back. It maps out targets on a palm computer. I did a short test, and concluded that GPS is not accurate enough for coin sized objects. At least civilian GPS. It was like a +/- sq meter.

          If you had a 1 meter coil and were looking for larger targets it might be very useful.

          http://www.magsurvey.com/MetLSurveyHome.htm

          Regards
          With a 1 meter coil, going in straight lines, no swinging, you would cover a good area and get repeatable results.
          The TINKERERS_TEM DETECTOR, can easily detect any coin or ring with a one meter coil.
          I would be interested to work with you on this project.

          Tinkerer

          Comment


          • #6
            Have you checked out the blanket coil ?
            http://www.youtube.com/watch?v=PYr9w...eature=related

            Comment


            • #7
              @bklein: No, it's not automated, it's a tele-operated, semi-automated system. What this means is I would be controlling the rover using the Xbox360 controller as a hand pendant. Think UAV drone command and control center. The operator (me) is sitting down in a comfortable chair, preferably with a beer next to me. I am watching the laptop, with the user interface and video being streamed from the Rover. When I get to the end of the scan, I would hit a single button on the controller, or screen, and the rover would do a perfect turn and shift over to the next area, all without having to do it manually. That, and the detector motion, is the semi-automated part. I agree completely that GPS alone isn't precise enough to just run as is. I would be using the GPS coupled with encoder position, and compass direction, to log indications as they are detected. I will mark out the perimeter of the scan area as accurately as I can, and use GPS plus dead reckoning of the Rover to get a good indicator. I am also kicking around the idea for some sort of triangulation method. I can buy a bluetooth Laser distance marker (Leica Disto D330i) for approx. $300 that has a 300ft range, and is accurate to a 1/2" or so. I buy three of these, each with there own GPS marker in each one. I lay out the 3 markers, and use that to increase my GPS resolution. For marking detected spots, I plan on using stakes for each indication seen, since I may not want to use the paint marker in all instances. If I do utilize the paint marker, say for when I'm using it in California, then I will simply have it automatically spray paint the location. As for the trouble to build such a rig, it's no trouble at all it's what I enjoy doing. I'm fortunate enough to actually design equipment very similar to this for my real job.

              @mario: I'm not too concerned with the 3G coverage. If I do have it, it would make the mapping easier, but it's not absolutely necessary. I will have a full blown map data set before I go to using it, just in case that's an issue. Although the rover will be nearly the size of a 4 wheeler, I don't intend on actually riding it. I intend to have "skid" steering, probably six axis for traction. This will allow for a very tight turning radius. The system will actually have multiple processors just because it's easier for me to build a distributed control system that way. I typically use various controllers to handle what they are good at, and not try to use a one size fits all mentality. I've been considering ROS, because it is actually designed to do just that, but I don't have any experience with it. My distributed control systems have all been to date a roll-your-own architecture. I would use the Propeller for localized, small control of sensors and misc hardware. I could do the sensor control from one of the other uC platforms, but I prefer the Propeller because of the vast user-created OBEX repository. Pretty much anything I would need has already be written by someone. Its a trivial exercise to take their objects and simply incorporate them into my code - usually without having to change anything other than the pin assignments. Plus its a true parallel processor, and having eight little cores to divy up the various sensors I want to use is quite easy. The PC I will be using on board will be running Flowstone, which gives me an easy way to interface to the Phidget hardware. I can have Flowstone running in my laptop, and have the Flowstone running in the Rover and simply functioning as a smart agregator and send the sensor data to the laptop. Flowstone has ethernet client / server modules, along with the Phidget and XBox360 controller modules, that allow for a really quick way to leverage PC hardware out there, such as the webcam. The Laptop computer is where I will also be recording a video stream served up from the from the USB webcam plugged into the Rover PC and I will overlay GPS and compass information, along with time and date, as well as metal detecting target information onto the video stream. I also may have some code running in Linux on the Phidget SBC, but that may not be necessary at first, since everything I need from the Phidget hardware can be accessed form the Flowstone on the PC. The nice thing about the webservice running the Phidget interface, is that it allows multiple points of control of the Phidget hardware. This allows me to be running the motors remote controlled, like a normally RC application, but in semi-autonomous mode, the Phidget SBC can take over and do local control as well. This allows for a safety watchdog estop circuit as well. If comms gets cut to the rover for a certain amount of time the SBC can independantly shut down the Rover movement without running off.

              @Tinkerer: I like how the mapping looks on the Garmin. However, I will be hopefully doing something on the Flowstone GUI interface on the laptop. It has a really good look I like, I'm even thinking of doing the control buttons, borders, etc, in a rustic retro look. The GUI will fill the whole screen, and will be just a compiled executable. As for the coil, do you have a link for your coil? It would be much better to not have to move. I had planned on scanning manually with a commerical detector, and I even bought one for the purpose, but not having to move would make things easier.

              @6666: Wow, that blanket coil is awesome! I really like that, I intended to have the detector on a similar dolly in front of the rover, but the idea of a blanket thats just rides behind the rover really appeals to me. That way I could just say, "ok go mark that spot" to my detecting compatriot, and its simplicity itself to go mark it with a stake or dig right there. Any information on this blanket coil like a link?

              Comment


              • #8
                Well, I've been doing some calculations on the power requirements. Looks like the rover will need a 5KW generator for purely electric driven drives. I may have to use the hydraulic drive idea after all such as the Prop QuadRover.. Here's an example of how powerful these motors are..

                http://www.youtube.com/watch?v=j1GK00oe170

                This will allow me to have the smaller 1kW generator

                Comment


                • #9
                  Since you have done this sort of thing before I just think you need more of a challenge. Your comparison to a UAV drone is right on. Don't bother with wheels - make the thing a quad copter. Those things are getting quite amazing. Might be a little dangerous if you are drinking beer though.
                  Or why not do the submersable? At least with that you can sit on your boat and drink your beer and not hurt anyone and get rich too. I guess GPS doesn't work underwater though... See, more of a challenge!

                  Comment


                  • #10
                    Building this with a quadcopter big enough to haul everything would require a quad copter the size of full size helicopter, not going to happen .

                    As to underwater submersibles, it could be done with a system that has an ROV platform with thrusters to allow for control along all 3 axis. The ROV could include video, along with sensors to assist in geolocation in conjunction with the GPS located on the barge. These sensors would include a compass heading, depth, length to barge info from tether length, any other information needed. This would allow for latitude/longitude offset correction from the barge to the underwater locations. One use of an ROV in this application would be to utilize a towable scanning array, similar to the blanket coil shown in the earlier post, that would be towed along the seabed floor, with the diver located just behind that with a tow pendant. Driving the ROV in a circular pattern around the barge using this technique might prove very productive.

                    I think using an ROV in this environment would require a different mindset on prospecting. This type of prospecting would need to be a precursor to a traditional dredging type operation. The diver could locate larger placer gold deposits, such as small nuggets and such, and any areas where larger gold pieces are located could be marked for further traditional dredging activities.

                    The easiest method to operating an ROV in this environment would be from the diver himself. His control, which would be built into a tow pendant, would have to be simple, waterproof, and intuitive to use. Feedback from the detector, located just in front of the diver on the floor, would have to be audible to both him and topside, and visual through the back of the ROV, so that he could note promising areas. Good two-way communications to the surface would be a must, so that he could mark areas as they are located with a stake. Scanning could be done with the diver simply driving the ROV as it tows him and the array. The barge would simply be needed to provide power to the ROV, and air and hot water to the diver, with the Rover providing the working storage platform for the diver as material is found.

                    There is a much larger amount of costs associated with designing hardware to run underwater, it would probably be much higher than the I can even guess. It all basically boils down to how sophisticated you want to get.

                    As to the Bering sea gold stuff. In all honesty, I would love to be able to go up there to do that. I've watched the TV show on Discovery. Who wouldn't want to be able to succeed at that? It looks good on TV, but they don't show the hell those folks have to go through daily to get what they do, and the winters they have to endure for a 3 month dredging season. Equipment failures and poor barge design seem to me to be to biggest showstopper, especially with the scrapping by mentality most of the dredgers seem to have deal with. Improve equipment uptime, and keep and retain a crew with a vested interest in taking care of the equipment, would be a game changer.

                    To build a ROV that could hold up to the environment up there wouldn't be cheap. ROV thrusters, enlcosures, etc, all would have to be sealed and pressure rated to at least 100ft+. Seawater I would imagine would also play havoc with that hardware. The fact that there isn't a large, established industrial base in Nome would also have to be factored in. The more complex and sophisticated the hardware is , the better the shop facilities need to be to order to maintain it. You'd have to have facilities sufficient to support your maintenance needs on your own, such as a full machine shop.

                    As to dredge design, one thing I see from a design point would be improvements to the sluice box design. Designing a detector into that, and redesign how the gold capture works, would probably make a big difference on the gold salvaged per hour. Moving the suction hose location based on feedback from the detector would allow for more efficient dredging. Feedback from the detector system to the hose operator would allow him to adjust to areas with greater signal feedback. The largest dredge on the show utilizes a backhoe dumping into a hopper. Removing the bucket and attaching a hose assembly would allow for an operator to control the hose based on feedback from the detector, and a camera mounted on the hose suction inlet. A few changes like that would probably provide a substantial improvement. Not having to remove the arm constantly back and forth, would allow for continuous suction. Something like that would allow the operator to work from a sealed cabin or inside the boat itself.

                    That's pretty much all I got on the whole Bering Sea concept. For this Rover project, I'd like to stay on topic.

                    Maybe after I build the desert/beach prospecting Rover, I could build a underwater ROV system. Who knows? It would be fun.

                    Comment


                    • #11
                      Accuracy with the GPS is going to be a challange for sure...

                      If you must use GPS to gather your coordinates then may I suggest you use a differential system?

                      This requires that you have a second GPS enabled unit located nearby at a location you absolutely know the true coords for. You then take readings at this location at regular intervals and log the data.

                      When the rover returns, you then upload the data and adjust it based on the time variable error that your reference has recorded.

                      Underwater isn't as tough as you might think as long as you engineer for the pressure. Maybe try fresh water first I know of a potentially awesome location in the great lakes region :-)

                      Comment


                      • #12
                        Once we get to the underwater project, I can apply my experience in offshore dredging for diamonds in Namibia. But lets start with the beach rover.

                        For the vehicle you mention, we could use a 0.5m x 1m coil. It will not pick up gold chains and very small jewelry but all coins, Rolex watches etc. However, the usual popular beaches are very littered. In many places the beach authorities contract companies which clean the beach litter every morning in the early hours.

                        I would suggest to combine a beach litter cleaning machine with the detecting machine. this way you have no problems with the authorities and get paid for the work. Cashflow is important. With the beach litter cleaner removing most of the bottle caps and other litter, you will have much less trash holes to dig.

                        There also would be a market for such a machine.

                        Tinkerer

                        Comment


                        • #13
                          I'll start a topic on the quadcopter and search for one on the submersable.

                          Comment


                          • #14
                            Great idea Tinkerer!

                            Can we have the "like or "thumbs up" feature activated please moderator? :-)

                            Comment


                            • #15
                              Originally posted by Farside View Post
                              Accuracy with the GPS is going to be a challange for sure...

                              If you must use GPS to gather your coordinates then may I suggest you use a differential system?

                              This requires that you have a second GPS enabled unit located nearby at a location you absolutely know the true coords for. You then take readings at this location at regular intervals and log the data.

                              When the rover returns, you then upload the data and adjust it based on the time variable error that your reference has recorded.

                              Underwater isn't as tough as you might think as long as you engineer for the pressure. Maybe try fresh water first I know of a potentially awesome location in the great lakes region :-)
                              Instead of GPS you could always use triangulation. Simply use 3 stationary antennas and use signal delays to determine better location coordinates. Civilian GPS will get better in the next couple of years, but will never be good enough for this type of application.



                              Originally posted by bklein
                              I'll start a topic on the quadcopter and search for one in the submersable.
                              Some good links on submersibles have been posted elsewhere in the forum. Two good links : http://www.homebuiltrovs.com/seafox.html and http://www.rollette.com/rovrev2/index.html

                              The current problem with submersibles involves communications. Currently, not many options exist when it comes to underwater, wireless communications. Salt water properties make it very difficult for transmissions, so something like wireless video feedback is currently impossible.

                              All home-made (and commercial ) submersible ROVs require physical data links (wire/ optical fiber) which limit their use greatly. Check out this article on a possible solution : http://m.popsci.com/science/article/...hrough-bedrock

                              Comment

                              Working...
                              X