Designing the Machine Learning board on the EDGE… BRAINCRAFT! @tensorflow @adafruit #machinelearning #tinyML
The idea (video) behind the upcoming BRAINCRAFT board (stand-alone, and Pi “hat”) is that you’d be able to “craft brains” for Machine Learning on the EDGE, with Microcontrollers & Microcomputers. On ASK AN ENGINEER, our founder & engineer chatted with Pete Warden, the technical lead of the mobile, embedded TensorFlow Group on Google’s Brain team about what would be ideal for a board like this. What did we come up with? Here is a start!
Inputs:
- Loads of sensors!
- More sensors can be connected via plug-n-play stemmaQT
- Image sensors, like a camera, or ‘image sensors’ like the Panasonic Grid-EYE. Heat-Sensitive cameras can be used as a person detector, even in the dark!
- Microphone, voice detection for powering on/off
- Accelerometer for gestures, vibration sensing. Could be useful for machinery/industrial predictive maintenance.
Outputs:
- Screen display for debugging
- Speaker for audio feedback as to what is going on.
- Relay control to turn things on/off, etc. and/or other actuators.
Connectivity:
- NB / Narrowband IoT? Wireless flexibility: WiFi, cellular, Bluetooth LE
- Works with adafruit.io, of course!
Power:
- Battery powered / low-power.
- Solar add-ons or energy harvesting. Make OK with lots of power sources that are not reliable.
Are we missing anything? Post up in the comments on what you want in a low-cost machine learning board! brAIncraft!
The name BrainCraft came from a company I worked at in the late 90’s that was acquired, around that time, I went off to Fallon, BrainCraft.com and everything about it, eventually went away, so I picked up the domain name about 10 years ago and have been using it for crafty projects, and now, finally, machine learning hardware. It’s kinda cool it all worked out, including have AI in the name, as well as craft and brain, since that’s how I think the training of models works for machine learning, it’s way more “craft” at the moment.
