Does anybody know of any good open-source solutions for habit box monitoring (e.g. bird boxes, bat boxes, hedgehog boxes and kestrel boxes)? I’m guessing movement (PIR or break-beam) sensors would be the simplest?
I’ve found a few solutions using a camera as part of a box design, but not any that can be added to existing habitat boxes. These boxes don’t have access to wifi or power.
Hi @gerrit!
I did some work in this space years ago, and PIR is the most common solution I’ve seen to trigger anything else you’d want to trigger (e.g. take a photo, record sound, etc.). Indeed break-beam is another solution but I’ve seen less of it compared to PIR.
In my experience, power is by far the biggest bottleneck, and you’ll need to do some work to make the PIR as low-power as possible. Some people use solar panels or big battery packs (whether that’s 12 or 16 AA batteries or one giant lead-acid battery).
Maybe @briannaljohns or @hikinghack can speak to the power challenge from their Mothbox experience?
I haven’t tried this, but a dry contact LoRa or Sigfox transmitter connected to a pressure plate switch might work.
LoRa and Sigfox devices that a triggered by the switch to send data and then go back to sleep can run off AA batteries for months or years.
@gerrit Alternatively, some optical detection are possible. These are some of ideas I used over the years.
Using inexpensive SJ3000 (often seen as SJCAM SJ3000/SJ-3000) action camera with a USB power banks, these have builtin motion detection.
Off the shelf CCTV, camera based on INGO chip allows you to load custom firmware using the thingino project. These chips have motion detection on chip. Power is provided via power bank.
USB web cam with extra long USB cable (2-4m max) connected to mini-computer running Yolov8 with detection limited to specific coco class for bird or mammal.
If you have Raspberry Pi with Camera and battery pack then you can run “motion” program for detection. See MotionEyesOS
Hoping to try newer low-power ESP32 and other boards including RPI AI camera to see if we could do on edge object detection.
I guess it depends what your budget for the box is and what kind of quality you need. Like you might be able to whip up something nice and low power with a PIR, cheap pi, and a little camera (and i bet folks on wildlabs have done this).
but like i have a friend here who uses a proprietary system that needs to take very good photos of bats flying quickly into a roost and it uses a ring of fast IR diodes doing a break-beam kinda thing connected to a DSLR that takes photos quick.
a nice thing about an actual DSLR is they are quite power efficient and can lay dormant quite some time but then be triggered easily to take nice photos. but that generally ups the cost of your system to at least like $350.
Visual-only sensing can be pretty dang power hungry.
I’d reckon that some folks on wildlabs make things like this mostly already, and if you have a bit more specifics on what the exact habitat boxes you are working with are like, there might be some good solutions popping up!
That’s why I was hoping there are existing solutions that have solved the power issue and why I don’t think a camera-based solution will work in my case.
Ooh, I like this idea. I was thinking LoRa, but still need to figure out if that can cover the range or whether we’ll need something like NB-IoT. Not sure where to look for pressure plate switches though, and will they work for bat boxes?
Yeah, I’m also hoping there are solutions with low-power boards. RPi seems overkill for just detecting if the boxes are being used or not.
We’re looking for low-cost options just to see if the boxes are being used or not. Sounds like I should get in touch with the Wildlabs community!
Agree with @hikinghack that WILDLABS is a good place to ask!
Hi Gerrit,
If you only need to know that the box is being used and you don’t mind physically checking on the boxes periodically, then I agree that PIR is the way to go because the sensor itself is very low power (less than 1mA for sure). The problem is an appropriate data logger isn’t available AFAIK. It’s not hard to make one though and I would give it a bash if I didn’t have to retool my entire so-called lab.
In case anyone wants to give it a shot, what I’m thinking is the sensor can be RE200B or DS203S or other 3pin TO can PIR, which works like a JFET, or HC SR501 module which has conditioned output. At any rate the output wakes a sleeping MCU, which logs the time stamp on its flash. Download the data via USB. Power with 2 AAs.
Good luck
Thanks for the sensor suggestions @Harold , that sounds relatively simple. Although I’m surprised that there’s not an existing solution already?
nth-ing the Wildlabs suggestion, the other place I’d wonder about looking is the ESPHome community. They’d likely be short-range options for comms though, and might expect power.
There seems to be a trend towards using the mmWave sensors for presence there, rather than PIR. I don’t know how they stack up power-consumption-wise.
I built something similar years ago but had trouble (more due to my electronics knowledge at the time) switching the cheap off-the-shelf PIR modules to 3.3V operation and so the signal was terribly noisy.
@amcewen I also wondered if there are any off-the-shelf LoRaWAN PIR sensors, but they all seem to be made for indoor applications. I suspect that’s going to be the case for ESPHome devices as well.
mmWave is an interesting idea!
I found these PIR breakout sensor boards from Sparkfun that have very low standby current consumption, and can then wake up the microcontroller to communicate over I2C (so they take care of the noisy analogue signal) - not cheap though: