embedded-research

The results of the undergraduate research student

index

1-ICM20948-ESP32-Raspberry_Pi_4B

• ESP32 get acceleration and gyro sensor value from ICM20948 9-axis sensor using i2c communication
• before measurement, calibrating acceleration and gyro value (using first N values from starting)
• raspberry pi get acceleration and gyro value from ESP32 using bluetooth(ble)
• raspberry pi can pause the receving the values.

2-anomaly_detection

• anomaly detection on raspberry pi using LSTM autoencoder model(using tensorflow and tensorflow-lite)
• training data and anomaly data have generated using getTrainingData.py
  • data is saved as .csv file
  • esp32 send the values, and raspberry pi process the data and save the data.
• LSTM autoencoder model is generated with anomaly_detection.ipynb

3-anomaly_detection_on_esp32

• anomaly detection on esp32 using CNN autoencoder model(using tensorflow-lite)
• using esp32-tflite-micro
• measure task: measuring sensor values and sending the values to 'detection task'
• detection task: detecting anomalies from receving values from 'measure task'
  • measure task send values using queue

4-anomaly_detection_on_esp32_ver2

• anomaly detection on esp32 using Depthwise-CNN autoencoder model(using tensorflow-lite)
• using esp32-tflite-micro
• measure task: measuring sensor values and sending the values to 'detection task'
• detection task: detecting anomalies from receving values from 'measure task'
  • measure task send values using queue