Project Summary

Languages: Python 3, C++
Tech: Sklearn, WeMos, MQTT
Timeline: Aug 2022 - Nov 2022
Role: ML Engineer
Team Size: 5

Situation

The elderly have high fall risk and weaker immune systems due to their age. We wanted to alleviate the burden on their caretakers with tech. So we created a smart walking stick to help monitor our them. It is loaded with features like fall detection and analytics on the cloud.

What I did

I created ML models for Fall detection and Severity. Using a MPU9250, I obtained 6-DOF (accelerometer and gyroscope) measurements of various falls. Then I carefully did feature selection from this data and trained a RandomForestClassifier and RandomForestRegressor to achieve an accuracy 99.0% with the model running directly on a WeMos R1 D2.