1. The raspberry pi waits for signal from PIR
2. When PIR is high [triggered], acquire images e.g. 5x images
3. Send these to TensorFlow for classification
4. If Hedgehog has > .80 then save these and upload them via 3G/4G or WiFi to twitter and to the hedgehog spotter website. Append GPS coordinates to the image metadata.
PIR circuit: that's just send 5v, and returns high/low to input
Light sensor: that's just send 5v, and returns high/low to input
IR illumination: we switch this by sending high/low from an output. Since it is 12v, and the Pi is 5v we use optocoupler (e.g. Isocom 4N25 DC Input Transistor Output Optocoupler, Through Hole, 6-Pin PDIP £0.20) , and we need transistor too.
Can we put all these onto a HAT? I guess so!
We also need a stepdown circuit for the 12v battery to 5v for the Pi. Can we put this on the HAT too? Like AmpZero did, so we power the Pi via GPIO.