Detecting and monitoring employee falls with embedded ML

Sure industries depend on staff with the ability to attain excessive areas by way of the usage of ladders or cell standing platforms. And due to their potential hazard if a fall had been to happen, Roni Bandini had the thought to create an built-in system that may detect a fall and report it robotically throughout all kinds of eventualities.

A fall will be sensed by measuring adjustments in acceleration; due to this fact, Bandini went with an Arduino Nano 33 BLE Sense board as a consequence of its built-in three-axis accelerometer. It additionally helps low-power consumption, that means {that a} LiPo battery and accompanying TP4056 charging module might be added for fully wi-fi operation. Acceleration information was collected by taking a number of samples inside the Edge Impulse Studio and labeling them both “fall” or “stand” when no motion is current. As soon as examined, the ensuing mannequin was built-in into an Arduino sketch, which emits a Bluetooth® promoting packet every time a fall is detected.

Accumulating every of those packets is the duty of a central Raspberry Pi server. It runs a Python script that always scans for brand new BLE promoting information and inserts a brand new report into its database file accordingly. All of this information can then be queried in a separate script and used to create a chart showcasing what number of instances each employee has fallen.

Extra particulars will be discovered in Bandini’s challenge write-up and Edge Impulse’s weblog put up right here.