This project implemented an activity recognition system using a hat fitted with sensors. The hat can recognize walking and running. This was the course project for "Measurement and Control Systems" course.
The hat uses the following hardware:
- Arduino UNO
- MPU-6050 accelerometer/gyroscope module
- MicroSD card module
- Battery pack
- Wires and connectors
The Arduino UNO collects the acceleration data from the MPU-6050 and logs it to a MicroSD card using the MicroSD module.
The acceleration data was collected only for the z-axis, as it provided the clearest signal for recognizing vertical movements like walking, running, ascending and descending stairs. Plots of this data over time can differentiate the activities well.
Sample z-axis acceleration data collected for walking and running is provided in the data folder in CSV format.
The Arduino code in the arduino folder reads the MPU-6050 z-axis acceleration value and logs it to the SD card.
The MATLAB code in the matlab folder loads the recorded acceleration data and trains a logistic regression classifier to recognize the activities. The trained classifier model is saved to disk.
The test cases in the test folder contain sample z-axis acceleration data for testing the classifier.