NoLoPhe / sensorstream

:compass: Visualize attitude tracking algorithm using mobile phone sensors

Visualize mobile phone sensor data with Python

Table of Contents

• Summary
• Introduction
• Instructions
  • Requirements
    • Client side (laptop computer)
    • Server side (mobile phone)
• References

Summary

Stream and visualize mobile phone sensor data in Python. As an example algorithm, a Quaternion-free inclination tracking Kalman filter.

Introduction

Mobile phones contain many interesting sensors that can be useful tools when developing e.g. sensor fusion algorithms for drones. Python is useful for quickly experimenting with sensor data processing algorithms before implementing the embedded version. This project contains an example implementation of a pipeline consisting of

• Data receiving from mobile phone,
• Sensor fusion algorithm for estimating sensor orientation,
• Real-time visualization on PC.

The inclination (gravitation) tracking in based on a quaternion-free method for estimating gravitation direction in sensor's coordinates. Visualization using OpenGL and Pygame.

Instructions

Requirements

Client side (laptop computer)

Python programming environment with the following additional packages

• PyGame: pip install -U pygame
• PyOpenGL: pip install -U PyOpenGL
• Numpy: pip install numpy
• PyKalman: pip install pykalman

Server side (mobile phone)

Currently tested only with the Android app SensorStreamer that sets up a mobile server for sending sensor data.

1. Install the app
2. Configure a data package with gyroscope and accelerometer data
3. Configure a connection with your favorite port (e.g. 3400)
4. Find out our mobile phone IP address (Search "IP Address")
5. Start a stream in the app with Lowest possible period
6. Run python sensorstreamer.py --host=123.456.78.90 --port=1234 --buffer=8192 --method=naive

References

S. Särkkä et. al, Adaptive Kalman filtering and smoothing for gravitation tracking in mobile systems