Unmanned Gold Detector Vehicle

Ordered the vehicle platform yesterday. Got the 12v kids ride on jeep with remote control. 4x4, handles up to 50kg load, and battery last for about 1.5 hours. Good enough for testing.

And I've taken a pic and video of the software so far. The video is a bit blury you can't really see the screen, so I've pasted what it displays below. Each new line is output every 10 milliseconds.

I've programmed my software to emit a "beep" sound every time it hears what it thinks is a signal. It emits a different pitch beep depending on the signal level, the higher the pitch the louder the signal. Every time you hear a beep, it means the software has logged the current GPS co-ordinates. Working well so far.





Max peak (16 samples): 0x00000335 # 2%
Max peak (16 samples): 0x000003e8 # 3%
Max peak (16 samples): 0x000006e8 ## 5%
Max peak (16 samples): 0x000004f3 # 3%
Max peak (16 samples): 0x00000599 # 4%
Max peak (16 samples): 0x000006a1 ## 5%
Max peak (16 samples): 0x00000593 # 4%
Max peak (16 samples): 0x0000056b # 4%
Max peak (16 samples): 0x000009c4 ## 7%
Max peak (16 samples): 0x00000696 ## 5%
Max peak (16 samples): 0x00000a29 ## 7%
Max peak (16 samples): 0x000047e0 ############ 56% ***** SIGNAL DETECTED. GPS CO-ORDINATES LOGGED TO /home/goldrover/gpscoords.txt *****
Max peak (16 samples): 0x00004247 ########### 51% ***** SIGNAL DETECTED. GPS CO-ORDINATES LOGGED TO /home/goldrover/gpscoords.txt *****
Max peak (16 samples): 0x00001afa ##### 21%
Max peak (16 samples): 0x000030d5 ######## 38%
Max peak (16 samples): 0x00004392 ########### 52% ***** SIGNAL DETECTED. GPS CO-ORDINATES LOGGED TO /home/goldrover/gpscoords.txt *****
Max peak (16 samples): 0x000049fc ############ 57% ***** SIGNAL DETECTED. GPS CO-ORDINATES LOGGED TO /home/goldrover/gpscoords.txt *****
Max peak (16 samples): 0x0000427b ########### 51% ***** SIGNAL DETECTED. GPS CO-ORDINATES LOGGED TO /home/goldrover/gpscoords.txt *****
Max peak (16 samples): 0x00007b96 #################### 96% ***** SIGNAL DETECTED. GPS CO-ORDINATES LOGGED TO /home/goldrover/gpscoords.txt *****
Max peak (16 samples): 0x00006680 ################# 80% ***** SIGNAL DETECTED. GPS CO-ORDINATES LOGGED TO /home/goldrover/gpscoords.txt *****
Max peak (16 samples): 0x00001ce0 ##### 22%
Max peak (16 samples): 0x00007502 ################### 91% ***** SIGNAL DETECTED. GPS CO-ORDINATES LOGGED TO /home/goldrover/gpscoords.txt *****
Max peak (16 samples): 0x00004e28 ############# 61% ***** SIGNAL DETECTED. GPS CO-ORDINATES LOGGED TO /home/goldrover/gpscoords.txt *****
Max peak (16 samples): 0x00005e1f ############### 73% ***** SIGNAL DETECTED. GPS CO-ORDINATES LOGGED TO /home/goldrover/gpscoords.txt *****
Max peak (16 samples): 0x00001aca ##### 20%
Max peak (16 samples): 0x0000657f ################ 79% ***** SIGNAL DETECTED. GPS CO-ORDINATES LOGGED TO /home/goldrover/gpscoords.txt *****
Max peak (16 samples): 0x00005745 ############## 68% ***** SIGNAL DETECTED. GPS CO-ORDINATES LOGGED TO /home/goldrover/gpscoords.txt *****
Max peak (16 samples): 0x00000a29 ## 7%
Max peak (16 samples): 0x0000051b # 3%
Max peak (16 samples): 0x0000008e # 0%
Max peak (16 samples): 0x000003d6 # 2%
Max peak (16 samples): 0x000003a0 # 2%
Max peak (16 samples): 0x00000335 # 2%
Max peak (16 samples): 0x000003e8 # 3%
Max peak (16 samples): 0x000006e8 ## 5%
Max peak (16 samples): 0x000004f3 # 3%
Max peak (16 samples): 0x00000599 # 4%
Max peak (16 samples): 0x000006a1 ## 5%
Max peak (16 samples): 0x00000593 # 4%
Max peak (16 samples): 0x0000056b # 4%

So is it measuring , Amplitude , Frequency of the audio or a combination of both ?

Hey Brad. Did you sort out the problem over coil interference in regards with the running motors, gps steering, onboard computer! Love the idea & where all waiting!
Cheers
Chris.

Eh Brad, I'm wondering if you can incorporate a device into your platform that can tell you if the signal is ferrous or not. Should be pretty simple and will save you a heap of false point visits.
cheers
Brett

Harb wrote:So is it measuring , Amplitude , Frequency of the audio or a combination of both ?

Neither, it's just measuring volume at the moment (in decibels). Pitch detection is next (low peak to high peak amplitude).

I plan on using an existing open-source oscilloscope software package that was written to tune guitars to assist with this.

someday wrote:Hey Brad. Did you sort out the problem over coil interference in regards with the running motors, gps steering, onboard computer! Love the idea & where all waiting!
Cheers
Chris.

Not yet as the vehicle platform hasn't arrived yet.

I've tested it with a smaller 7.2v brushless DC motor and there were no interference problems with that at 1 meter from the coil.

But the motors on the new vehicle I've bought are bigger, so still need to test.

The computer, GPS receiver and digital compass are fine - no interference from them.

mulgadansa wrote:Eh Brad, I'm wondering if you can incorporate a device into your platform that can tell you if the signal is ferrous or not. Should be pretty simple and will save you a heap of false point visits.
cheers
Brett

Hey Brett - yep, that should be fairly easy to do I think and is on the to-do list. I still need to properly work out how the "Iron Reject" setting works on my GPX-4000 and what "blanking" sounds like, I haven't played with it much yet I've always had it set to "OFF" while I've been learning.

In the GPX-4000 manual It's got this picture. The one on the left is when Iron Reject is off, and the one on the right is when Iron Rejcect is on:



Could someone who uses a GPX-4000 tell me what "blanking" actually sounds like. Is it a silent gap in the audio like it shows in that graph? Do you hear background noise or nothing at all?

In that video of that guy finding the 44 gram nugget recently, I know he's using a SD2000 but he describes "blanking" as that loud screaming sound so I am a bit confused.

Hi Brad, great homework program your working on, should be good when you get it up and running. I'm afraid it's a bit over my knowledge as I'm not much for electronics.

On the blanking, I use a 4000 and with a DD the target starts and then when right over it the threshold just blanks out (the audio goes dead for that second) which indicates that it is ferrous but, as the the Manual states you need to make sure that the rubbish is not masking another signal. I have found that on a Mono it will do it on large, loud targets but it is hardly noticable and so far they have all been rubbish. I haven't used a DD for sometime but run discriminator on 2 with the Mono only because it seems to run smoother.

Cheers bergs

Gday Brad
Don't use the discriminate on the GPX - totally useless!!!
cheers
Brett

mulgadansa wrote:Gday Brad
Don't use the discriminate on the GPX - totally useless!!!
cheers
Brett

Without using the Iron Reject setting on the GPX, how could my software work out from the audio signal if a target is ferrous or non-ferrous? Are there any differences in high-low or pitch that it might be able to identify?

bergs wrote:Hi Brad, great homework program your working on, should be good when you get it up and running. I'm afraid it's a bit over my knowledge as I'm not much for electronics.

On the blanking, I use a 4000 and with a DD the target starts and then when right over it the threshold just blanks out (the audio goes dead for that second) which indicates that it is ferrous but, as the the Manual states you need to make sure that the rubbish is not masking another signal. I have found that on a Mono it will do it on large, loud targets but it is hardly noticable and so far they have all been rubbish. I haven't used a DD for sometime but run discriminator on 2 with the Mono only because it seems to run smoother.

Cheers bergs

Thanks Bergs, I just went and tested it out and now understand what you're saying. Cheers.

Gday Brad and Bergs
My understanding was that the disc didn't work at all with a mono coil. I haven't tried it.
I have tried using the discriminate on my 4500 in heavy rubbish areas in old townsites and it is useless. Sometimes it will blank on ferrous material and sometimes not. Don't know if I'd trust any info coming out of an "Iron Reject" setting if I was searching for big gold with the type of platform you envisage.
What I was saying in my earlier post was that I thought you may be able to incorporate some other ferrous indicating device into the platform
cheers
Brett

brad28b wrote:

Harb wrote:So is it measuring , Amplitude , Frequency of the audio or a combination of both ?

Neither, it's just measuring volume at the moment (in decibels). Pitch detection is next (low peak to high peak amplitude).

I plan on using an existing open-source oscilloscope software package that was written to tune guitars to assist with this.

Ok , so what you are currently measuring is amplitude, which of course won't discriminate..... it will react to anything that changes the "Volume" unless you set an aplitude limit or "Squelch" arrangement.

So what you really have to do is data log not only amplitude but also frequency of the sound emmitted, but even better still, set up commands to make the device, stop reverse , recheck the signal and if it is present still, drop a waypoint or a paint marker at that spot.... that will cancel out atmospheric bursts.
If you really wanted to perfect it, you could have it compare the Freq/Amp relationship to known data that could be sampled from targets that you would set up and capture data, and then make it give a percentage of probability that the target is gold, and maybe even upgrade the waypoint recorded or spray a different color on the ones that go over a certain percentage.

What you really need is a Audio Spectrum Analyser software or USB package (they will measure both amplitude and frequency )to mark windows of probability...... much better than an ocilliscope that only measures Amplitude and timing..

Once you perfect all that, you could then add a Parimetric Equaliser to all the "gold" sounds through at a much higher amplitude than anything else, further isolating gold only sounds.

Hey mate I've got a spectrum analyser package already - been playing with it the last few days. But at the moment my own software isn't using it:

http://www.baudline.com/screenshots.html

It will be eventually

Sounds like a Plan

A lot of Software packages have the ability for the setting of "alarms" if certain perameters are met..... keep an eye out for that function....its exactly what you need.... they will then cause a reaction...like sound an alarm or switch something via a GPI trigger etc etc

People can tell the "Gold" sound........... so an instrument like that will have no trouble matching the sounds when a true hit is made.

Harb wrote:Sounds like a Plan

A lot of Software packages have the ability for the setting of "alarms" if certain perameters are met..... keep an eye out for that function....its exactly what you need.... they will then cause a reaction...like sound an alarm or switch something via a GPI trigger etc etc

People can tell the "Gold" sound........... so an instrument like that will have no trouble matching the sounds when a true hit is made.

Yep baudline can do that, you can set a threshold, then set a "high" limit to trigger an alarm to do something (which in my case will eventually trigger a script to log the GPS Co-ordinates).

Thanks for your help also it's good to be steered in the right direction, I've only found 1 piece of gold so far so im not very familiar with the good sounds yet Becoming an expert on the junk sounds though.

Good luck with it...... I will wait to see you on "New Inventors" or on the Ford list of rich people

Don't hold your breath

Well if it ends up not working on gold, you will be able to sunbake on the beach while it finds you coins, watches and jewelry

cheers

brad - what harb is saying - about people being able to "tell" its gold and not ferrous, is true, to a degree.

from my experience, the easily heard signal can definately be identified by the operator as possibly gold. it is a softer and sweeter sound, as compared to the harsher sound of metal. as harb describes, your fancy equipment - which is all dutch to me - should be able to tell the difference.
but when you get to the faint sounds, it is a whole different ballgame, not at all obvious, and this is where i feel a whole heap of improvements could be made to existing machines to give the operator a more certain idea that the target is in fact gold.(without resorting to DDs and discrimination mode)

most detectorites have been swayed towards mono coils in recent years, as they punch deeper and are more sensitive, which suit the newer machines, and therefore, nobody uses the discrimation button much (wont work with monos).

crack this one and you will be on a winner. best of luck ... kev

G'day Brad,

I hate being the party pooper, but there is no way to discern whether a target is gold/non ferrous or ferrous from the threshold. Put a 30 ounce nugget 3 inches from the surface and you have,.... steel junk! Try a nut at 12" sweet mellow sound, gold!

The discriminator only works with a DD coil. It is taking a late sample during the tx on time where the signal amplitude will be higher for ferrous, but highly magnetic ground interferes with this measurement causing the discriminator to be unreliable.

Cheers Mick

hehehe
very nice project there!
the original post could have been written by me- had exactly the same idea (coming from the same background lol )
would i be living in australia, i could probably offer some support... but from the other side of the world (and no equipment) it's a bit complicated
but good luck!!
cheers

Hi brad,

If you can’t solve the vehicle for detecting purpose don’t give up on it. I once was dreaming up on a small support vehicle following me in a constant distance of 2 meters, stops when I stop, don’t move when I get near to it. The vehicle carries all the stuff I need: Water bottle, gps, battery, pick, folding shovel and of course my sandwiches ect., (Payload 5kg). Anything you otherwise have to carry on you and might interfere with the coil readings and works on you by the constant weight.

If you could come up with something reliable I believe you make many friends. Especially with those who like to swing themselves and spend hours or days out in the bush.

Cheers
ALL-AU

All-AU wrote:Hi brad,

If you can’t solve the vehicle for detecting purpose don’t give up on it. I once was dreaming up on a small support vehicle following me in a constant distance of 2 meters, stops when I stop, don’t move when I get near to it. The vehicle carries all the stuff I need: Water bottle, gps, battery, pick, folding shovel and of course my sandwiches ect., (Payload 5kg). Anything you otherwise have to carry on you and might interfere with the coil readings and works on you by the constant weight.

If you could come up with something reliable I believe you make many friends. Especially with those who like to swing themselves and spend hours or days out in the bush.

Cheers
ALL-AU

You can buy a golf cart that will do all that you just may need to modify the way it carries things, but it follows along via a radio link.....

hi mechanic - have to agree with you that "knowing" a signal is gold is well nigh impossible per sae.

the reason i though brads equipment might be able to improve on human ears is because of a recent experience : - detecting with a 5000 and NF mono 17x11 coil, i had found more than five ounces in a area that was close to a very filthy dump area. (biggest being 28grams down to point somethings) the 28 grammer was down about 500mm and as i was in reverse signal mode i got a nice sweet mellow sound.

not far away was a sunbaker of 22grams, so loud, i felt for sure it had to be a piece of wire on the surfact, it was a different sound to surface tin, of which there was plenty around, along with many inbetween sizes and depths. after figuring i had got them all (and we all know nobody does), i decided to tackle the filthy area.

turning the volume down a lot lower, i was able to find a four grammer, several 1grammers, in the ground, and then was driven crazy by all the continuous metal sounds for what seemed an eternity. i had practically got into "zombie mode", wasnt checking all these obvious(?) metal signals, when a signal pulled me up and it turned out to be a 20gramm sunbaker.
for some reason that i cannot confirm/identify/understand, my brain had process what my ears had heard, and told me, "that was different".
i didnt say - BS - and throw it away.
in my world, nothing is impossible, so i look forward to the day when someone like brad, improves on this technology ... kev

brad28b wrote:

bergs wrote:Hi Brad, great homework program your working on, should be good when you get it up and running. I'm afraid it's a bit over my knowledge as I'm not much for electronics.

On the blanking, I use a 4000 and with a DD the target starts and then when right over it the threshold just blanks out (the audio goes dead for that second) which indicates that it is ferrous but, as the the Manual states you need to make sure that the rubbish is not masking another signal. I have found that on a Mono it will do it on large, loud targets but it is hardly noticable and so far they have all been rubbish. I haven't used a DD for sometime but run discriminator on 2 with the Mono only because it seems to run smoother.

Cheers bergs

Thanks Bergs, I just went and tested it out and now understand what you're saying. Cheers.

Hi again Brad, I PM'd you earlier this month on this matter and you seem to understand the blanking sig etc...etc..,now, and if you still insist on sound analisis? you Probably could use a bit of lead or gold to get the spectrum you are looking for. But I told you that I had a very simple method of testing for ferrus or non ferrus but it involves taking a sample of the TX sig and the RX sig. This can be done by a small box that would sit between the coil and the control box. The coil would plug in the bottom end and a patch cable would go from the top end into the control box. This could have a red led for junk and a green led for good-non ferrus(or center-zeroed meter). It would be able to be used by the many ($300) detectors out there already, that dont have this facility. But you could use the internal facilities of the unit to assist your softwere. You have my No so If you get stuck with the direction your going, ring me or PM me for further info. But your doing a great job so far so keep it up
regards
Tony
Tony

Keep that up Tony. you'll fall into the same category I'm in! Trust me you don't want that? Well??Maybe??Narr?? Up to you?? Don't say I didn't warn you!

someday wrote:Keep that up Tony. you'll fall into the same category I'm in! Trust me you don't want that? Well??Maybe??Narr?? Up to you?? Don't say I didn't warn you!

Nahhhh Chris I think he's a good kid and means well, I think he knows his **** and it will be good for the forum to see some more progress and coments happening.
regards
tony

I agree mate "Rome wasn't built in a day" But if it's personal, keep it to P.m's Trust me I know the ramification's and it ain't pretty!

Thanks Tony, I do remember you sending me that, I will just need to work out a way to incorporate something like that into the platform. Sorry for the lack of updates, been flat out with work. Does your device have any outputs other than LEDs that I could hook into with my software (eg Serial, TTL, USB etc?) Or could it be modified easily to incorporate that?

The test vehicle platform arrived yesterday. It's bloody big. But its got plenty of grunt to tow the gear. Time will tell how suitable it well be. I've also received most of the electrical parts that I need to do the conversion. The steering I'll need to order some more parts for, but building a motor driver should be pretty easy by the look of it (it's got 4 x 6V DC motors, one on each wheel). And a steering motor (I think it might be a linear actuator) which I still need to work out the specs on. Need to remove that red shell first, and replace it with a sheet of plywood to mount everything too. Also need to replace the tyres with rubber tyres. Should have time next week to get started.

Before I get too far into the autonomous navigation, im going to take it out and do some testing just using remote control, to test the electrical interference from the motors, vehicle capabilities etc

Here is a quick video of my dogs welcoming it.



By the way its got 2 speeds, high (7km/hour) and low (3.5km/hour), this video was in high speed.