Placemon consists of a number of different sensors:
- Microphone
- Temperature
- Humidity
- Pressure
- Light
- PIR
The idea is to use the data collected from these sensors along with machine learning techniques to recognise different events occurring in your home. For example processing audio data, to see if you have left a tap running.
One of the key inspirations is the paper - "Synthetic Sensors: Towards General-Purpose Sensing", I noticed some of the sensors used alone on their board such as the GridEye thermal sensor appear to be pretty expensive. I've tried to select a smaller set of sensors that will hopefully allow us to achieve a similar level of accuracy at detecting different events.
I am planning to capture the data from the sensors locally, using a single board computer, such as a Raspberry Pi, to perform the machine learning on the data obtained.
Schematic
I need to add a pull-up resistor on the GPIO0 pin.
Render (with incorrect esp32 footprint currently)
PCBs
Component orientation
I created a little script that loads the KiCad centroid file that contains positions and rotations of components, and checks the position of pin 1 and generates a figure which you can see below (the white rectangles represent pin 1).
With this script, you specify the KiCad centroid file and libraries as command line arguments, so that appropriate libraries can be loaded to determine pin positions etc.
https://github.com/anfractuosity/pcba
How many pcb layers?