Driver for MPU6050 IMU Sensor, and Pose Estimation with Kalman Filter

MPU6050 is a low-cost IMU sensor, which provides 3-axis acceleration and 3-axis angular velocity measurements (6 DOF in total). This sensor is really handy when it comes to implementing things like wearable devices and drones. I got an evaluation module of this sensor, and decided to build a very simple interface for it using a 8-bit MCU. Source code is available at https://github.com/YifanJiangPolyU/MPU6050

A video showing how it works:

MPU6050 has build-in ADC and is capable of outputting readings digitally via a I2C interface. In this interface, a Arduino Nano, which has a 8MHz 8-bit MCU, is used as the I2C master. The Arduino reads measurements from MPU6050 cyclically, and estimates the pose of the sensor, represented as Euler angles, using a Kalman filter.

The estimated Euler angles are then transmitted to the PC via serial port (i.e. UART). On the PC side, a python script reads the pose data and visualizes it in real time.

The current implementation has a number of problems, but improvements are also easy to implement.

1. The 8-bit MCU is not powerful enough to execute pose estimation.
2. Delay is sever, because of inadequate processing power of the MCU.
3. Communication protocol between Arduino and PC is too primitive. Better protocol can be used to enable data integrity checking and miss data detection.
4. The data acquisition and the Kalman filter run in an infinite loop in main(). A better way would be to make this interrupt-triggered, for a more deterministic sample time.