I'm trying to build what is essentially an underwater camera trap. The basic premise is you have a camera pointed at something, and you want the camera to take a pic when something comes along. In this case we'd be putting the camera maybe 20-50cm from the subject. The one project that seems to have successfully done this uses an array of red LEDs, shining on a subject with the camera constantly on, with a Pi connected, analysing when the image is significantly different (ie. something is there), which triggers a strobe and takes the picture. This is both too expensive and too complex for what we're looking for. The fallback is just having a time-lapse camera, but it'd be good to do something novel and more effective.
Sensor requirements:
-needs to detect whether something is coming between the camera and the substrate. I don't need to know the actual distance, just that it's different.
-needs to still work at 20m depth. I can make housings to deal with this, but if the sensor is running a membrane or similar, I assume 3ATA of pressure is going to cause problems.
-cheap. I've got the camera control side of things sorted. I know I could buy a $300 ROV sensor to do it, but that's not viable.
I've read up a bit, and it seems that the typical acoustic sensors are probably out. Even if I can make them work, the extra dead space created by the speed of sound in water is going to cause problems. The question is whether the JSN-SR04T is worth trying? I've read a lot of mixed reviews on this.
Otherwise, something visual. Water absorbing IR is going to cause some problems, but given the short distance it's a matter of whether it's worth potentially potting an IR sensor with optically clear epoxy? I've got plenty of the Sharp IR sensors laying around. I also have some of the Laserping sensors. Putting them right up against the glass of a full aquarium leads to it giving a zero reading, but I don't know if this is because it's hitting the glass and bouncing off (in which case putting it in a housing as-is is out), or it hits the water and bounces off. Again, I could pot it if it might work? I gather it's emitting near-IR.
Led of appropriate wavelength to propagate through water. Sensor for led. When something comes between sensor and led there will be a drop off in light detected. Activate camera
Or echo sounder if you don’t want sensors on either side of the subject. Use a speaker to make sound signals and a microphone to pick them up. Send out signal and detect return signal. There are off the shelf solutions for this but I’m sure you could bodge something.
No. Underwater sonar transducers are completely different than those intended for operation in air. Most, if not all commercial sonar rigs are intended for distances greater than about 1 m.
For 20-50 cm, a reflective sensor using blue light should work. IR and red light is strongly absorbed by seawater, which calls into question the projects you mentioned.
You could try replacing the IR LED and phototransistor on one of the simple reflective sensor modules with a blue LED and a visible light phototransistor, but for underwater use, everything would have to be extremely carefully potted.
Thanks, that's a dead end I don't need to investigate then.
It's a matter of how quickly IR and red are absorbed. There's still some red in visible light at 20m depth. Camera strobes light up pretty big areas with no noticeable loss (look at the spectra in photoshop or similar). Some UW cameras use a red light as a focus light (to not spook fish).
This is something I definitely should have mentioned. Whatever I use it needs to have a negligible effect on the fish behaviour. There's plenty of examples of fish not caring about a camera, but if the camera has lights and makes lots of sound, they will care. People use red lights because fish are less likely be able to see it. They're much more likely to be able to see white or blue lights. UW animals often love light, and will be attracted to it, or hate it, and will run away from it. Something noisy, or having lights visible to them, is going to necessitate an experiment testing whether it affects fish behaviour, which we'd like to avoid.
I'm guessing you want coverage over an area, not just a single spot (or line is more correct). I don't know how quickly a fish could sense a light but you could use a red emitter (as mentioned above) and modulate it so its only on for a short period of time.... say a 500µs every second, interaction with the fishes** might be minimal.
** yea I know one fish two fish red fish blue fish.
Fair question. Partly because the whole thing will be run through an arduino of some sort, but primarily because this forum is the best source of information about building cost-effective solutions involving sensors I've found on the net. Hobbyists and ecologists seem to have the same motivations: either find a cheaper solution to something someone else has done in an expensive or overly-involved way, or find a solution to a problem no one else has done yet.
Users here, you in particular, have steered me away from my ignorant impulses a few times, and it's something I really appreciate. I can deal with the ecology side, but I'm still very much learning about arduino, and sensors in particular. Users here have a breadth of experience that googling or a lit-review can't replace.