AI Wearable Companion, 2018

The idea for Project Oco was born from two hypothesis:

1) Current mental wellness products fail to address some or many user needs. For example, a wellness product should:

  1. Be always accessible

  2. Facilitate emotional interactions

  3. Demonstrate understanding

  4. Provide actionable help

  5. Support wellness outside of a crisis

2) Current AI companions have "bodies" - or physical forms - which can be improved to better support human interaction. Better bodies should:

  1. Be mobile

  2. Provide sensors with which to perceive the world and understand human emotions

  3. Enable physical interactions with the user

With 2 months and $300 from MIT ProjX, Magnus Johnson, Milo Knowles, and I sought to create a conceptual prototype which addressed these two hypothesis. Our result is Oco, a wearable focused not on notifications and physical health, but on companionship and mental wellness.

Oco consists of a Raspberry Pi, ePaper display, accelerometer, pulse sensor, gesture sensor, and camera, as well as solenoids, vibrating motors, and batteries in a wearable form-factor. The Raspberry Pi sends facial images from Oco to Google's Cloud Vision API, which can provide mood predictions. This information is combined with biometric and physical data to estimate the emotional state of the user. Oco can communicate to its user through means of naturalistic taps via solenoids. Oco also presents itself visually in the form of faces rendered on a low-power ePaper display.