Ripple v0.1.0

First public pre-release.

Added

• Motion and presence detection from the RSSI of the link to the router: a 2-second
  sliding-window variance against a calibrated threshold, with hysteresis and a
  30-second presence hold.
• A guided calibration (empty room, walking, sitting still, empty room again) that
  derives the threshold from the room's own data and reports a quality verdict.
  Quick calibration and a manual threshold are available too. Calibration is saved in
  flash.
• A web dashboard served by the board: live signal chart, status, calibration,
  sensitivity, MQTT settings and device info. Light and dark themes, works on a
  phone. It has a built-in demo mode with a simulated board.
• MQTT with Home Assistant discovery: motion, presence, signal strength, signal
  fluctuation, a motion threshold you can change and a calibration button, with
  availability and Last Will.
• WiFi setup through a captive portal (Ripple-xxxxxx), a ripple-xxxxxx.local name,
  and a way to forget the saved network.
• REST and WebSocket interfaces, serial commands, and a project website with a live
  demo and a web installer.
• Unit tests for the algorithm, the calibration, the Home Assistant messages and the
  WiFi helpers, and continuous integration.

Known limitations

• No dashboard password and no signed over-the-air updates yet. Update over USB.
• The presence timeout is fixed at 30 seconds.
• A person sitting perfectly still cannot be detected with the RSSI alone.

Install

Flash from your browser with the web installer,
or write ripple-<board>-v0.1.0.bin at offset 0x0 with esptool.
Check the files against SHA256SUMS. Pick the image for your board: nodemcuv2, d1_mini or esp12e.