PDA

View Full Version : [Question(s)] How to detect / find plastic objects outside



J4S0N
03-25-2015, 10:33 AM
Hello everyone! I'm a bit new to this and find myself getting in waaaay over my head... I want to build a device that can scan and find plastic discs outside but I can't figure out what type of sensors I need or if it is even possible...

My requirements are:


Cheap! I don't have much money to spend on this
Able to scan at least 10 feet away (but 30 feet or longer would be a big plus)
Able to identify 6 inch diameter hard plastic discs in any orientation (flat on the ground, or at any angle)
Able to scan behind/through tall grass, bushes and tress
Scan cone can be slim, assume can be spinning like a radar
Small/Portable design, hand held would be great



My first idea was passive UHF RFID tags on the disc but the price of the readers are too far out of my budget. I prefer not to attach things to the discs but can if it makes finding them easier (must be light weight and very slim - like a sticker).

My next thought was sonar or ultrasound, but i'm unsure if it can detect plastic and if it can look through bushes. I've only now started looking at this type of sensor. Is that the right path or is there something else I should be researching?

Thanks for reading, I hope you guys can point me in the right direction!

tician
03-25-2015, 03:24 PM
125kHZ RFID readers can be found for really cheap (<$15), but the range is usually limited to a few inches at most.

Unless the plastic is 'metal detector' rated (small amount of ferro-magnetic material added to the plastic during manufacture), then
there is little you can do to detect it other than using vision sensors like a cheap webcam. The vision sensor combined with appropriate processing hardware and software would only really be able to provide candidates for a disc if they were sufficiently distinct in color and/or shape from the background/surroundings, and would require further close-up examination to determine if actually a disk or just noise/garbage.

J4S0N
03-26-2015, 07:54 AM
Thanks for the reply. I looked in to LF and HF RFID and found the range too limiting. UHF was the only one I found with the range I need but the price is too high for the readers. I like the idea of each disc having an ID but it's not a requirement at this time. Locating the discs in the woods is the goal.

I started looking at OpenCV and doing vision but there are cases where the disc will be covered by leaves or behind trees or bushes where vision will fail. If I can't see, I doubt the device would see it.

How much ferromagnetic material would I have to add to be detectable and what sensors would I need? Would the metal in UFH tags be enough? I think it's copper but would have to check. Sorry if that is a dumb question I haven't really researched metal detection yet, but I like your idea!

tician
03-26-2015, 11:07 PM
Metal detector grade plastics are primarily for things like food processing equipment where any plastic components breaking off the equipment might otherwise pass through QA and into packaging undetected. Food processors usually have metal detectors somewhere on the line prior to packaging to ensure no metal tools or contaminants make it through. Some also use gamma, X-ray, and/or UV/Vis/NIR for finding things like bone in meat and performing some forms of quality testing and disease/contamination detection, but those tools are usually expensive, potentially dangerous, and/or probably not very effective in your particular situation.

If the plastic has special spectral properties (greatly absorbs or reflects some range of light) not really shared by environmental objects (plants, rocks, bricks, etc.), then you can use that to aid detection by illuminating the scene and searching for unusual patches of dark/light. Big box stores usually care small rolls of retroreflective tape, so you could stick a few patches onto the disks to make them stand out a lot more when shining a spotlight (especially if the color of the spotlight is one that is readily absorbed by the local foliage).

jwatte
03-27-2015, 10:24 AM
there are cases where the disc will be covered by leaves or behind trees or bushes where vision will fail

It sounds like your requirements and your budget are not in sync, then. If the thing will be covered, you need long-range RFID or something like it.

What are the discs used for? Skeet shooting? Can you put an active beacon of some sort on the discs, and use triangulation? Will they be visible from above; can you use a quad copter to find them?

J4S0N
03-27-2015, 10:48 AM
The discs are used in frisbee golf. The idea of the device is to find my disc when I throw it in the woods.

Weight is a big factor and limits what I can attach to each disc. My first prototype was using BLE "stickers" but the cost was $20 each and they were a bit big. While the range of bluetooth was good, measuring distance based on signal strength left a lot to be desired... I couldn't find a good way to say the disc is to the left or right.

I'm not against active tags but they have to be very small and no more then a few grams in weight.

I agree long range UHF RFID is a great choice if budget wasn't a factor. I've looked at building a UHF reader but would still need to invest $400-$500 and it didn't appear to be very portable. I don't need the ID part, if there was a simple reader that just beeped and pointed me to the disc I would be set. Are there any other radio type devices I could look in to?

Is there no passive way to locate a plastic item? If I added a small amount of metal, how much would I need to detect it from a few feet away? (the further away the better). I don't know much about metal detectors, but I will start researching it.

I assume I could build some type of sensor that is tuned to locate some specific metal. What would be the best choice? (keep in mind weight is important).

Is Ultra sound or sonar even an option?

tician
03-27-2015, 12:29 PM
Metal detection is often achieved with a coil of wire as the inductor in a resonant circuit. When a metal object is placed in sufficient proximity to the coil, the EM field of the coil interacts with the metal object and changes the inductance of the coil which shifts the resonant frequency of the circuit. They can be small like hand-held metal detectors or large like the coils embedded in the road for vehicle detection at 'not completely brain dead' lighted intersections. Detection distance is still usually limited to a few inches due to limited power of the coil.

You could use RF and/or sonar as in an old-school animal tracking collar. Basically, just a small RF/sonar transmitter producing a known output (e.g. a repeating sequence of pulses on a fixed transmitter frequency) attached to the object to be tracked and a directional antenna/receiver to identify the direction to travel.

Plenty of detailed descriptions and schematics online for all of the above.

jwatte
03-27-2015, 02:01 PM
Sonar is great, because two microphones can give you direction (through phase difference, if you turn it around) and four microphones can give you specific direction.

long-distance RFID is poor for this use case because it typically requires multiple sensors placed around an area, rather than just carried.

The main question is going to be how much you're going to accept in weight (and cost) for the actual disc device. It's not like this is a new idea. Golfers have the same problem with golf balls, and usually are more willing to spend money than disc golfers. If there was an easy and cheap way to do this, it would likely already be available as a product!

J4S0N
03-27-2015, 02:33 PM
So it sounds like passive detection at the range I want is not doable :-(

I really want to find a way to do this without needing to attach transmitters on each disc. I'm not only worried about weight, but these things will be thrown at high speeds and smacking into trees and the ground.

Is there any way to use RFID tags without the need of an expensive UHF reader? From what I understand the reader sends a signal that powers the tag and the tag then responds back with its ID and the reader reads it. I assume I can't just build something this blasting out radio waves and an antenna looking for what ever the tag is sending back? I don't understand enough about RF to get to that point myself.. but maybe??

Adding transmitters would only be doable if they were very small and cheap... I'd still prefer to go the passive route, but that is starting to seem impossible (i hate that word)

I did some very basic research on RF tracking and there seems to be a few products that are cheap but are kind of big. Saw some on amazon called Click N Dig which might work but I'd like to find something much much smaller:

http://www.amazon.com/s/?ie=UTF8&keywords=click+%27n+dig

I might poke around the web some more this weekend see what kind of DIY transmitters I can find. But I'm questions my ability to make them rugged enough to be thrown around on a disc.

Any advice/links/projects/other resources you guys can share would be much appreciated!!

J4S0N
03-27-2015, 02:49 PM
@jwatte I agree! If this was easy it would be done and I'd go to the store and pick it up!

Can you tell me more about how I could use sonar? I assume it's simple to measure distance based on time of travel but can it detect different materials? Can it tell the difference between the disc and a bush? Can I still find a disc if it's partially covered? Do you have any links to projects using for different objects?

I can be somewhat flexible on the requirements. A typical disc weights between 150 and 175 Grams, I don't want to add more then 25 grams to that. Form factor needs to be slim since there is only a small lip (maybe 1/4" to 1/2" ) on the under side the disc for this to fit in to.

Cost requirement is tricky, I'd like to spend under $100 total but would be willing to spend more as needed.

tician
03-27-2015, 03:22 PM
jwatte was describing multilateration of an active transmitter using multiple microphones and signal processing (measuring time difference between arrival of wavefront at each microphone; more difficult than it sounds if not using an easily distinguished sound/frequency and/or many sources of noise, distortion, echoes, etc.) to detect its direction and distance from the microphones/receiver. i.e. the tracking collar using sound instead of RF. Could be as simple as a lithium coin cell and a tiny PCB with a cheap microcontroller and piezo speaker to make a chirp at a fixed rate after some length of time without its snooze button being pressed.

jwatte
03-27-2015, 06:56 PM
Yes, I was assuming an active transmitter on the disc, and using a microphone array to triangulate phase. A chirp with a known frequency at a known interval is necessary to tell signal from surround.

There are small and cheap acceleration sensors. PCBs can be made very thin -- even flexible, depending on specifics (I'd recommend rigid.) There are small and cheap piezo chirpers, lithium batteries (or supercaps!) and 8-pin microcontrollers. You can get this into a few grams and half a square inch or less. Make the thing only turn on and chirp after it's seen acceleration and then stopped -- so it's laying on the ground -- and time out after a few minutes to save battery/charge.

Is $100 enough? It'll probably cost you more than that just to make the first set of PCBs, and the solder mask for soldering the surface-mount parts to make it small and light. And you'll likely need to re-do it a few times to get it right.
Also, microphones, and multi-channel sound input, and a CPU + direction indicator output, will likely be another $100+ (I'd go for a Raspberry Pi 2 and a USB stereo sound input for simplicity)

J4S0N
03-28-2015, 09:52 AM
If I have to go the active transmitter route, I think i will need a lot help... I write software for a living but I am very much a n00b when it comes to designing/building electronics. Thanks to my high school electronics class 15 years ago i know how to read basic schematics and hold a soldering iron with out burning myself lol.. I'm rusty but ambitious :-)

Would anyone be willing to partner up and help me design something like this? I doubt I can make something that will be small enough...

I'm still very interested in finding a passive way of detection. I might start looking into active RFID as well.

J4S0N
03-28-2015, 04:24 PM
I spent too much time looking at uhf rfid stuff again... i found a few projects doing things like what i want but most have faded out and never got completed. I think to make this portable i need to accept a smaller read range and raise my budget. here are a few things i found, thoughts?

http://www.edaboard.com/thread200506.html
http://ams.com/eng/Products/UHF-RFID
http://www.ebay.com/itm/Cheap-Mini-Size-UHF-RFID-Reader-Module-for-Access-Control-/111631527104

J4S0N
04-01-2015, 11:53 AM
Anyone? Thoughts?

jwatte
04-01-2015, 11:59 PM
Let's re-think this problem. You're looking for a disc in some field/woods/thickets.
Discs, alone, tend to be kind-of hard to spot, even if you get them in neon-ish colors.

However, the eye is good at spotting glimmering lights. And the environment of the disc ought to reflect/amplify some light source on the disc.

Thus, the simplest possible solution to "how do I find my disc" is to get a coin cell battery, a small trickle-flashing circuit, and a high-intensity LED. And perhaps a very lightweight switch, if you don't want to use "insert battery" to turn on the flash. Mount this to the center of the disc. The disc will now make a flashing light, perhaps once every second, which should make it much easier to find. With the right brief flashing period, and flashing interval, the coin cell battery will last a long time (although you want to turn it off when not playing.)

Xevel
04-02-2015, 12:46 PM
Do that, and add a chirping buzzer so that you can get a general idea of the location even if it's hidden in a bush or under leaves. Also helps in area with a lot of contrast (lots of sun and some dark shadows from trees).

J4S0N
04-03-2015, 11:21 AM
I like your idea and I have been doing a ton of research. I am a complete n00b and have never designed anything on my own. I've spent hours on digikey trying to figure out what I need, but I'm lost.. so I looked up a few kits to try to figure this out on my own:

http://www.radioshack.com/radioshack-flashing-led-kits-pack-of-10/2770230.html - I Like this one but its out of stock :-(
http://www.radioshack.com/velleman-dual-white-led-stroboscope-kit/2770101.html - way too big
http://www.radioshack.com/velleman-mk102-flashing-leds-kit/2770231.html - also too big

Can someone give me a hand? kick start me with a parts list or something? I want to learn and build something but I'm in over my head.

On a side note, I came across this "Simple Low Cost UHF RFID Reader": http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.362.3294&rep=rep1&type=pdf

They built a basic reader but the read range isn't that great. Any thoughts on how it could improve the read range to at lest a few feet?

birdman
06-30-2015, 09:17 AM
I think it -is-possible passively. The key is (as others have pointed out) a matter of selectivity--you need something that 1. Presents a strong signal when interrogated, even if only a small portion has LOS. 2. Is different signal than anything else (as much as possible)

i would suggest a NIR illuminator (IR LED array) with a narrow band, and then a cheap little board cam with a matched IR bandpass filter (cheap, filters can be had for $10-20). Then put retroreflectove tape on the disk. If you want even better SNRS, cover the retro tape with NIR transparent (but blocks visible) film--that way only your interrogating light gets retroreflective.

that should work at really good range, even a small area can be readily detected (retroreflective >>> just reflective, as its 1/(2r)^2 intensity instead of 1/r^4), and the sensor / illuminator can be really cheap.

it doesn't address situations where there isn't a LOS, but then the periodic acoustic beep could work, as Xevel said,

Do that, and add a chirping buzzer so that you can get a general idea of the location even if it's hidden in a bush or under leaves. Also helps in area with a lot of contrast (lots of sun and some dark shadows from trees).

but I wouldn't bother with auto-detection of when to beep (accelerometer, etc) I would just make it active when you use it...a short (<50ms) beep every 10s coukd use an extremely small amount of power (microW average), can be built using low cost parts (small piezo disk buzzer small SMT discretes, home-made flex, and a really small LiPo or coin cell). For start/stop make a latching circuit that latches on/off when two contacts show lower R (with a finger?) or a very easy flex-circuit based switch (have a folding tab with a bit of tape that you can tape closed or open circuit depending which way you fold it).

for RX, two microphones (be wary, most mics dramatically lose sensitivity above 15-20kHz, unless they are really expensive) record signal, either analog-filter or digital filter a bandpass around a high resonant freq of the piezo buzzer, then centroid each signal's envelope, and do a TDOA calc to determine azimuth. Moving a short baseline (hence the 10second interval) and repeat should give okay range estimate.