Beach Detecting Rover project

[fixstuff]

Well, I'm shooting for automating something that we do manually with a hand detector. I personally will sweep a detector a couple times standing still, move a bit forward and repeat. I don't walk and scan very much. Maybe most people do it that way, but I guess I like taking it slow so I don't miss anything.

That's how I plan on scanning with the rover. Rover stops, arm sweeps once to the left. Arm extends out 1/8 or 1/4 inch. Arm sweeps right. Repeat until arm stroke is reached. Move arm back to beginning. While arm is coming back, move rover forward 12". Stop. Repeat whole process.

This way the rover isn't adding to arm bounce. In addition, I will have one of these mounted to the arm boom section.

http://www.maxbotix.com/products/HRLV.htm

I'm going to use it to keep the arm height constant, by adjusting the arm riser. I'm even thinking of having a bank of them (say 5 or 6) in the front of the rover pointing down, and mapping the terrain height as I go. Maybe eventually colorize terrain height and compare it with the Kinect point cloud map. That's some of the long term projects I will test out.

If I position the ultrasonic sensor well enough, I can use it to prevent the coil from hitting anything, say monitor average height, and if I see something that suddenly jumps up, I can shut down the coil sweep.

[fixstuff]

Detector arm assembly work

Hi folks,

Worked on my detector arm this weekend. Below is the final picture of the rotation base I made for the arm. It doesn't look like much, but it pretty much occupied most of my Sunday. There is more info and pictures on the base parts on my blog if anyone is interested.

http://fixstuff-thots.blogspot.com/

[fixstuff]

new memorial weekend work

Hi all,

I made some good progress this weekend on the rover. I redid the arm, redid the rotation, and mounted the non-driven, dead reckoning encoder wheels. I also did some metal detecting testing of the sweep arm. It looks very promising. I have more info on the blog here:

http://fixstuff-thots.blogspot.com/

Here also are some links for the video I took of the testing.

http://youtu.be/lhHwaLNyYWM
http://youtu.be/ahc5_83dn40
http://youtu.be/6guV0vBTPLs
http://youtu.be/jUmfdfXMpiA

I've also attached some pictures.

[fixstuff]

work this weekend

Hey folks,

Well, I don't have pictures today, but I wanted to give a status update. I've got all of my electronics parts mounted today. I stepped back, and re-did the first panel I mounted a couple weeks back because I didn't like the screws sticking out of the lid. I got another box lid, and basically put everything on a plate. That way only 4 screws that are symmetrical stick out of the lid. Looks way better, and is less like to leak also. I had to buy aluminum for that though. Yuck. Aluminum plate is EXPENSIVE. I bought 3 12x16 3/16 aluminum plates and 6 12x6 plates and it was over $120. Ouch. Anyways, it looks very nice now. All the electronics are mounted, using nylon insert jam nuts, and its looking very nice. I have even wired in the 120V mains input wiring from the generator.

As it stands here is the electronics mounted:

1 HTPC Zbox computer
Intel CPU - Atom - 2 GB RAM - 1.8 GHz - With HDMI - NVIDIA GPU

1 XMOS motor controller board
1000 MIPS XS1-L2 Two-core programmable device
RJ45 100M Ethernet socket and PHY
CAN 2.0B interface via DE9 connector
37 Digital I/Os
6 analog inputs into isolated dual 1MSPS 12b ADC
4 user LEDs and 4 push buttons
128 x 32 graphics monochrome LCD display
50-way IDC header containing power, digital and analogue IO for user expansion
512KByte SPI boot FLASH
XMOS Link and XSYS debug interface supporting JTAG and XScope

1 Phidget SBC2 computer
Single board computer running Debian with access to the full Debian package repository.
Web based configuration and monitoring interface.
Abundant on-board storage.
Run custom programs on-board for autonomous operation.
6 full-speed USB ports with built in support for webcams, Phidget USB devices
Integrated PhidgetInterfaceKit 8/8/8.
8 digital inputs, 8 digital outputs, 8 analog inputs
Ethernet port. Remotely use attached Phidgets from another computer anywhere over the network.
Realtime clock with battery backup.

1 Phidget 4ch high speed encoder board
4 high speed encoders
4 digital inputs
1 Phidget GPS module
1 Phidget 3 axis Gyroscope /3 axis accelerometer /compass module
1 DSP Robotics Flowboard
16 digital inputs
16 digital outputs
8 analog inputs 10bit
4 Pololu 24V23 motor drivers
3 Phidget Single Motor Driver boards
2 digital inputs x 3
2 analog inputs x 3 12bit
1 encoder input x 3
4 24V, 14A power supplies
1 350W ATX power supply
1 Wireless Router with 4 ethernet ports
1 Phidget Text Display
8 digital inputs
8 digital outputs
8 analog outputs
2 line 20 character VFD

Rover:
1.85KW generator
4, 350W wheelchair motors
2, 75W linear actuators

As a side note, I found some really nice textured black paint I will paint the rover with. Not sure yet if I will paint the treadplate (mainly cause it looks so nice), but for the boxes and arm it will be great.

[der_fisherman]

Painting aluminium

As a side note, I found some really nice textured black paint I will paint the rover with. Not sure yet if I will paint the treadplate (mainly cause it looks so nice), but for the boxes and arm it will be great.

Don't forget that aluminium needs a special primer or the paint will flake off later.....
I always first clean the metal with methylated spirits or similar (it has different names depending on where you live in the world!), and don't touch the surface after cleaning or wear rubber gloves (clean/new ones).

Best of luck,

Andy

[fixstuff]

quick heads up

Just a quick heads up. Last night I ran my first motor on the rover! I ran the motor over the network using VNC and a webcam monitoring the motor. I did this with a static metal detecting test as well. It didn't appear to have any affect on the MD. Either with the motor off, on, full speed, or cycling the the ranges the md "blipped" about the same as it did with the motor power off completely. It wasn't a true proper test (4 motors simultaneously), but it does bode well for later.

This week I'll wire up the rest of the motors, and start testing them this weekend.

video link to the testing:

http://youtu.be/gKHBe1UvDTs

[fixstuff]

Motor testing

Well,

I got the other motors wired up. It looks like 3 work, and 1 doesn't. The motor is good, but the amplifier is dead. It came up for a couple minutes, but then the lights turned off. No smoke, not smell, just crib death amplifier. The wiring was good, but these little amplifiers are delicate, delicate, delicate. Not sure how long they will last.

Anyways, at least 3 work.. This tests out most of the power hardware.

I had to pull the ATX power supply. I only need 12V out of it, and I needed to space, so I killed it. I will use a dedicated small 12V supply to get the 12V.

Well that's all I got for now. More updates in a couple weeks..

Next step is replace the bad amplifier, and power up the Phidget hardware and Flowboard.

[fixstuff]

Amplifiers

Been doing some thinking since last night. From a system architecture perspective, my methodology for interfacing may need to change.

I had planned on running all of the motion control through the PC from the high level software control only. Based on what I'm seeing in terms of bootup time and general ease of use, I'm thinking of adding a subsystem hardware layer for motion control. Every amplifier I have can be controlled through an an analog interface as well. If I build a small "motion controller" board, then I can still control motion from the PC, but I can also run the motors without the PC, or before bootup. This will allow me to simplify the rover controls if needed. I can preposition the rover with an RC controller or something similar, without having to turn the PC on. Once the PC is on, I can perform the semi-automated scanning functions.

I will build the controller and mount it right on the amplifier drive lid inner panel so that the interfacing wiring is local and on the panel. I will have both a Zigbee interface for the board, and the USB interface so that the board can still accept commands from the PC, but also can be used when connected over a wireless connection.

[Farside]

...I had to pull the ATX power supply. I only need 12V out of it, and I needed to space, so I killed it. I will use a dedicated small 12V supply to get the 12V.

I don't know what your power requirements will be but this is a good cheap one:
http://www.dealextreme.com/p/12v-1a-...-94519?item=42

Shipping free :-)

Actually, there's all sorts of neat stuff here.

[Infamy]

Hi,

I've only had a brief look at your project so a few of my comments maybe a bit of track.

A detector which works on mineralized areas, is waterproof, has pots, good depth with some discrimination and can be bought for under 500 euros. A second had original version Fisher CZ20. They are a rugged piece of equipment, probably just what you need.

If your project is to only search beaches, and is not made up of way points that the machine progresses too automatically then I think you are really taking a one ton hammer to crack a walnut.

I can't see the point of a front mounted swinging coil, dragging it is less complicated and more efficient.


I use to work on GPS machine guidance on the coal mines in Australia mainly with Leica and Trimble system so I may have a few leads for you. The GPS problem you curently have will go away very soon with the ESA satelite system and the new GPS L5 signal which the USA is bring into service. Three gps receivers on your rover will not do much for you, currently you need to have one on a known position which corrections can be calculated from and then sent to the rover or you computer. This will change soon in Europe as two stationary satelites should do this, it may have allready happend.

I wrote this in 2009 there are some low costs high accuracy RTK solutions here. 2010 updates have been added.

Introduction to the Introduction

I started on this project about a year ago, scribbling a few bits and posting them on the net when I had time. What follows is a compilation of mostly stuff I have all ready posted. Some of its now out of date, where this is the case the current situation will be set out in Italics.

Autonomous Treasure Hunter Part 1

Introduction

A few years ago I became interested in treasure hunting, wondering if it was possible to change the odds of success. Is it possible to locate treasure using maths and science, is there a special formula? A common thread in the jigsaw is time coupled with research, but time is not a simple variable, nor is research, they are both multi dimensional. The utilization of time to physically hunt as many hours as possible will increase your chances of being successful, but is it the key to maximizing the use of time? Research can help to maximize the use of time, but are you sure its not being wasted by following in the tracks of others?


Autonomous Micro Survey Vessel

When I started on this project it was my intension to build an autonomous electric glider for aerial photography. On close examination of autonomous flight I came to the conclusion it was not going to be of sufficient benefit as I already use a standard RC glider for aerial photography. Instead I decided on constructing an autonomous boat with a sensor carrying barge. The aim is to search difficult places in shallow water that people look at and think, its too hard. Shallow water areas in Europe are interesting because many ancient settlements now lay just a few feet below the water surface. This is the result of several factors, a preference for settlements near the waters edge, parts of Europe sinking due to the ending of the Ice Age and rising sea level.


Precision Multi Layer Analysis (PMLA)

PMLA as described here at the time of writing refers to data that has a 3d shape in WGS84 coordinate system, where Z values are manipulated to represent the form of sensor information. The data can be analysed by using spreadsheets and digital terrain modelling software.

Data types of interest are Precision Proton Magnetometer (PPM) data and Pulse Induction (PI). A PI detector set to locate all metals, and a Precision Proton Magnetometer (PPM) only detecting iron. The theory is, the PPM data cancels out the iron signals in the PI data, leaving non ferrous targets, of about 1lb plus in size. (The theory here has been simplified, and I have taken a different route. Instead of using a PI detector I am using a Minelab Soreign GT which can give tone discrimination which I intend to log using a miniture USB sound recorder. This route also allows the Magnetometer element to be been kicked into the long grass,)


RTK GPS

In order to achieve PLMA the physical location of the data needs to be survey quality or the data will be too degraded to be of any benefit. Imagine trying to match data for different days which is a couple of meters out. At 2-5 cm accuracy 90% of the time the job should be clear cut, thats what RTK GPS should provide. A base station receiving GPS carrier phase information will calculate corrections, then transmit them to a roving receiver. RTK is the pivotal technology for collecting the data.


Open Source

When you have an idea such as building an autonomous vehicle how can you do it without spending large quantities of cash? go Open Souce (OS)

“Times they are a changin” Bob Dylan

This subject of OS is so dam crazy. Why give away ideas, code, programs, data, for free? Thats basically what is happening in the world of “Open Source”, its 80s free gaming utopia gone mad.

For me one of the benefits might be the advancement of cheap RTK GPS, in OS, technology moves on at a rapid pace and can bypass tecnological and anti competative barriers. It is however rather difficult to benefit from if you try to access information in areas where you do not already have the skills to understand, modify and implement the information. Much of the technical information is poorly written and barely passes for English. OS as it appears currently is not the final utopia in my opinion; fragmentation and diversity may be a great benefit, but it also means it has lots of small spheres of influence, which are not currently penetrating much beyond the geek world, Linux being the exception to the rule. What is needed is finished product and ideas which anyone can implement.

OS example
www.diydrones.com

Product development OS variant
http://www.gizmoforyou.com

Red Lake OS variant
www.troymedia.com/NewsBeats/Technology_News_Beat/2009/02/MCBD022709.htm

Technology Situation of Project

General

The PI detector I have set up for the first trails will be an Infinium by Garrett. Its waterproof, but the main draw back is the difficulties in making or acquiring coils greater than 14 inches. I intend to switch to a Whites TDI preferably a waterproof version if it is released. The benefits of the TDI is it can accept a wide rarity of coils and is suitable for DIY coils. ( But the Sovereign GT, costs less, has discrimination and it is feasible to build a 40 inch coil. Only negative point is it aint waterproof, need to be careful here)

Work on the boat structure is complete. A data logger, GPS receive, autopilot and PI detector have all been hooked and are under going testing. I'm waiting for the power unit and Kort Nozzle setup which will drive the boat. Its a system specifically for tugs, very efficient at low speeds.


RTK GPS

SweetGPS written by Steve Waslander was one of the first open source RTK softwares. As I assessed this software, a bomb was going off in the world of open source RTK GPS.
The Laboratory of Satellite Navigation at Tokyo University of Marine Science and Technology released an open source RTK suite with an English manual in September 2009.


T.Takasu and A.Yasuda the authors of the software have produced virtually everything that is needed to develop your very own RTK system
http://gpspp.sakura.ne.jp/rtklib/rtklib.htm

Paper on the development of low cost RTK
http://gpspp.sakura.ne.jp/paper2005/...tklib_revA.pdf

The low cost RTK project ($400)
http://gpspp.sakura.ne.jp/rtklib/rtklib_beagleboard.htm

Evaluation of L1 antennas and receivers
http://gpspp.sakura.ne.jp/anteva/anteva.htm

GPS Analysis Software
http://gpspp.sakura.ne.jp/gpstools/gt_release.ht

The downside is that current low cost technology is not at a point where a mobile vehicle can solely use RTK-GPS for precise position. The time when it will be possible is very close due to improving technology and the convergence of cheap IMU and GPS technology plus the commerical pressure and improved accuracy of alternative satelite networks.(The ideas about where cheap RTK GPS was going to come from have moved on, I think Nokia will release a mobile phone within the next 2 years which will give 100mm accuracy, if the price is cheap, 300 – 500 dollars this will be a major technological development)

The original version at DIY drones. http://diydrones.com/profiles/blogs/...reasure-hunter


I'm not an electronics expert so my solutions tend to be taking whats all ready there and plugging it into something else. I end up with a system that works but is not seemless

[fixstuff]

@Infamy: Thanks for the info, I really appreciate the feedback. I have always really intended to use the detector in other locations than beach use. I may tinker around with it on a beach one day, but it's not what I'm really shooting for. I'm really working on this rover as a desert or field detector. I'm also using it as a platform for other potential applications not related to metal detecting. Anything I develop for this rover, will help me out on my other applications.

For the detector, the sweep arm I'm going to try to run at a continuous rotation. This gives me a 15ft scan radius, and is easier to build. I set a challenge to myself to see if I could build the sweep arm, and so far it is looking very promising. The sweep arm with it's extendable boom will allow me to test a PI circuit with a coaxial spiral coil design, in order to detect very small gold deposits. Coils like this, from what I can see, are somewhat impractical when used manually due to the small detection area, but for this setup might be more suitable.

Whether I sweep or not, I am also going to try mounting my commercial metal detector to the arm. By doing this I can leverage good commercial detector tech, and I will simply transmit the audio over wireless straight to the operator and mapping hardware. For my Gamma 6K, I can cycle through different modes remotely to give me a better picture of what targets are made of. I'm also going to try out a GT2500.

If the sweep arm proves a bust, or if I can't run it do to objects or ground clutter, I may simply keep it in a fixed position, change over to a large coil, and simply use the rover in drive mode. The arm is also not my only plan of detecting, I have plans to build a towed phased array setup and I will use the PI kit I built earlier and use it on the towed array. I've found some good plastic yard seeders that should work great as a starter frame for that.

As far as open source goes, the project is really intended to help me realize it with advice from others such as yourself. I don't plan on commercializing the project, since I just don't really think folks are prepared to drop $10K+ on this. I truly think being open can result in a better end result, than keeping it hidden. I think metal detecting technology could really go through the roof if it wasn't hoarded so much. I hope to be able to do my part to help out with this.

The PMLA idea is kind of what I want to do with the data anyway. I planned on using Matlab or something similar to layer different sensor scans over the same mapping data. I planned on reading targets with different modes on the detector, and mixing the results into a color overlay on the map, which is similar to what you were talking about.

I was planning on overcoming most of the GPS issues with dead reckoning on the rover. Non-slip wheels with high resolution encoders should help a lot. I've read several papers that demonstrate GPS accuracy can be substantially increased by using multiple sensors, and averaging the over time from a fixed location. Not moving the rover except during arm extension retracts may be able to achieve this. I'd love to see more accurate GPS tech though. Hope it materializes.

Thanks for the links you provided, I will read up on them and see where they can help me out.

[knowthebird]

Fixstuff,

This is off topic Im sure as most of this looks over my head and indepth. Im interested in building a metal detecting robot for counter ied/mine detecting. Something cheap do it yourself, but can still navigate through moderately rough terrain. a sweeping lane about 3 feet wide (might be too much) but the vehicle still being able to be picked up and moved or carried if needed. My plan on was using some premaid chassis and servos for it as I have no machining tools or the money to have anything custom made controlled by one of the numerous motor micro controllers available. I feel fairly confident in my programing ability, and understanding of basic circuits, but im really a user when it comes to metal detectors, not a maker. Id appreciate any advice on where to get started with a reliable metal detector for this project to reach around 8"-1' foot depths in most soil (seems reasonable?) that i could incorporate into a micro-controller to relay to a rc unit. and any advice for a beginner trying to take on this project. also, if this is the wrong place for that discussion message me for my email address (dont want it getting spammed). thanks ahead for any and all help.

[Qiaozhi]

Fixstuff,

This is off topic Im sure as most of this looks over my head and indepth. Im interested in building a metal detecting robot for counter ied/mine detecting. Something cheap do it yourself, but can still navigate through moderately rough terrain. a sweeping lane about 3 feet wide (might be too much) but the vehicle still being able to be picked up and moved or carried if needed. My plan on was using some premaid chassis and servos for it as I have no machining tools or the money to have anything custom made controlled by one of the numerous motor micro controllers available. I feel fairly confident in my programing ability, and understanding of basic circuits, but im really a user when it comes to metal detectors, not a maker. Id appreciate any advice on where to get started with a reliable metal detector for this project to reach around 8"-1' foot depths in most soil (seems reasonable?) that i could incorporate into a micro-controller to relay to a rc unit. and any advice for a beginner trying to take on this project. also, if this is the wrong place for that discussion message me for my email address (dont want it getting spammed). thanks ahead for any and all help.

This stuff looks really interesting ->
http://www.valuframe.co.uk/
You could use it to build a custom frame without any complex machine tools.

[Pennypacker]

What ever happened with this project. At first I had a feeling he may not finish it, then it looked like he was, then he didn't lol.

[fixstuff]

What ever happened with this project. At first I had a feeling he may not finish it, then it looked like he was, then he didn't lol.

I still am working on it I've been busy on a couple big projects at work, so I work on it occasionally when I can..