I didn’t know where else to post this, since I’m actually discussing multiple issues at once. So this seemed the best place of all the forums.
I’ve had in mind a robotics project for some time now. Life sort of got in the way before I could start working on it, and good thing too because there’s been so many advancements in robitics and the associated field I want to apply it to. I love being outdoors and most of my hobbies are outdoors related. One such hobby that I’ve enjoyed now for over 10 years is metal detecting. Unfortunilly humans and accurately scanning the ground doesn’t always work out so well. Conditions that can as I see it be very easily over come by using a robot instead of a person.
A brief course in metal detecting…
Metal detecting is basically sending a very weak radio signal into the ground. Depending on what’s directly under the metal detector’s loop at the time, you’ll recieve a singal back which the modern metal detectors will then analize and then tell you roughly whats down there. There’s a few issues with this. First and formost is how that signal gets into the ground and how rapidly it dissapates. A standard 12" loop has a signal that’s 12" across at the surface of the ground. Depedning on ground moisture, minerlization and how sensitive you have your detector set, When you get down about 6-8" deep, that signal is now the width of a quarter. In order to be absolutly sure you’ve gone over every bit of target area, it’s best to overlap your sweeps by at 50%. A quarter is just shy of 1" across (24mm). That means you only want to move your loop forward half an inch (12mm) for you’re next sweep1 Sounds easy, but in practice it’s not all that easy to do. Physical fatigue, distractions from the enviroment around you, loss of concentration because you’re re-thinking of what a jerk your boss was to you last week, all that you you WILL miss targets in the ground.
To make matters even more difficult for the average detectorist, that 50% over lap doesn’t always guarente you’ll see a good target. That’ assuming that a coin is laying perfectly level and flat below ground. What if it’s on edge or at an angle. Worse yet, what if the ground is “trashy”. You’ll never find it unless you go over the same ground twice! Say the first time going north to south, then a second time east to west. (You could hypothetically say you need to approach the same field from all four directions, but that’s really going a bit far!) This way, if that target which is on a slight angle may not show up going one way, but has a larger surface area when approached from a 90* rotation in sweep.
Robotics to the rescue!
I’m sure you all can see where I’m going with this already. You’d have to be living in a cave not to have seen at least one news clip showing robots doing mine sweeping in either Afghanistan or Iraq. This is actually what gave me the idea in the first place! If a robot can find a land mine or IED, why not a silver coin, gold rings, or a nugget???
So what features do I need to work on to accomplish what I want my robot to do. This is where I need advice. ALOT of advice. And that’s why I’m here.
I’d like to do this in roughly three stages. The first of course will be a learning stage (all stages will be learning really, but this is more like testing/learning not applying/constructing the final unit) The second stage will be a working robot that I can go off and do some serious detecting. The third stage will be more advance with some complicated features that I’m not so sure can be done, but I’m willing to work on it!
Stage 1 (testing)
I think I can use something like Lynxmotion’s A4WD bot. This will be where I will test sensors and practice my programming skills. Sensors I’m going to need to use in order to accomplish what I want my bot to do will include 1)object detection and avoidence 2)ranging from “base” 3)differencial gps so I can preciely determine where a good target is from my “base” while the bot is in the field. 4)real time data telling me where the bot is and reporting what it’s detecting.
Before I reach stage 2 though there are a few things that I could use a hand with, even now. Not that I’m ready just this minute to being constuction, but I could use advice on certain design issues, and where I can get information so I can over come these issues. My first hurdle is size of bot. Several companies make a roving bot simlar to the A4WD. The problem is, they’re not large enough to carry a detector, battiers and such. There are a few companies that I’ve already found like Parallex makes that new gas powered unit, and The Machine Lab makes their larger MMP unit, but I’m not THAT wealthy to afford either of these! ($5000-$8000!!!) I do realize, it could eventually cost that much to build what I want to, but I cannot afford to drop that kind of money in one chunk.
My next hurdle to over come is the actually mechanism that I will use to sweep the metal detector’s loop back and forth. I’ve got two ideas in mind, both having pro’s and con’s. One is to use an arm that sweeps left and right, just like a human’s arm would swing a detector. The advantage here, is if I have the bot programed to automously detect a certain area in a random fashion so it’s free to avoid obsticals. Imagine gold nugget detecting someplace in Arizona or southern California where you’ve got sage brush, cactus, bolders, ect. Advantage here any sensors can be program to prevent the metal detector’s loop from hitting these things, but getting as close as possible without doing so. The disadvanage is the sensors themselves. Mount them on the chasis, the arm/loop will interfer with detection. Mount ANYTHING metal on or near the loop and you get interference and false target signals. The next concept I had was to rig some sort of carriage arrangment such as like your average laser printer moves the print heads back and forth. I could use a stepper motor to move the loop back and forth, but I won’t be able to get as close to obstructions as I would if I used an arm.
Such a carraige type arrangement has another distinct advantage that I could then use some distance sensors and a second set of steppers to raise and lower the loop making sure it’s kept just above the ground but without hitting it.
Of major importance is battery power. With all those sensors going, motors advancing the bot, steppers or an arm constantly in motion, and the sheer vastness of area that needs to be covered, small battery packs just won’t do. According to the specs on the MMP unit, fully charged they only have 1 hour’s useage before needing to be charged again. I’m thinking along the lines of using a law tractor battery or something simlar.
Distance and usablity. While at home, I take my bot to a local park or forest preserve, this will be the easiest detecting. Land is basically level. The drawback is people. Teens with nothing better to do, or people looking to make a quick buck would more then likely drag my bot off without hesitation. Therefore while on home turf, I’ll be in sight of the bot. But my ultimate goal is to hit the desert south west to do soem gold nugget shooting. I know this will be an issue and I haven’t even begun to research it all yet, but how do I maintain controll over my bot, if I send it out a mile from my “base camp”? At the very least I need to hear what the detector is detecting, I can then manually mark that on a map going by gps coordinates that I recieve as to where the bot is. Later I can use these markings as waypoints when I go out into the field with a second detector and my shovel.
I have all these issues worked out and a working bot that can go out into the field, then I’ve gotten past stage 2.
Stage 3.
From the get go, I’m going to use a dedicated laptop for all my programing and eventually field work for the bot. Hopefully by this time I will have found a robot club near me, because some features I would like to add from here on out will require some heavy duty programing and some hacking of the metal detector’s circuts. Fortunilly some of this has already been done for me! Since I first came up with this idea, a metal detector company in Austrailia called Minelab has come out with a detector that can be programmed via a usb port. If this could be exploited properly, I could potentially gain access to control the detector pod from the laptop in real time! This way I can make notes such as “way point 1, target detected at approximatly 4 inches deep” It’d also be nice if the bot could monitor the responce from the detector as well. That way it could automatically program the waypoint/targets to my laptop. All I would need to do then is download those waypoints into a portable gps. And just for yucks, I was tossing around the idea of having the bot somehow mark the spot where a targe is detected. Perhaps a gallon bottle of water with some bright blue food coloring and a solanoid sprays a shot of blue water over the spot. When I come along I go over that ground with my second detector, mix the soil up, and nobody will know I was ever there. and food coloring wouldn’t be harmful to the enviroment.
Anyways… this certainly will be a fun project to work on. It may even take me a year or three to actually accomplish. But there’s certainly worse things I could be doing with my money and time, right?
Besides reading threads here and elsewhere… What other reference materials should I be looking at? I’ve already found one book that I would LOVE to get immediatly and that’s “Robot builder’s Bonanza”. Just from the brief bits that I read while browsing around Border’s book store, I like the author’s style. Many of the other books have me scratching my head and wondering why I haven’t taken a class in quantium physics or rocket science!
Thanks,
Jax