Getting useful motion data from the MPU-6050 device

If you have played with motion sensors, like the MPU-6050 6-axis gyroscope and accelerometer device, you may have noticed that the raw data that you can read from them contain too much noise to be useful.

The values can fluctuate wildly, even when the device is sitting motionless on the floor.

This happens because the raw data that are generated by the device contains a significant amount of noise.

Clearly, using the raw data is not an option. There's not much you can do with them other than printing them to the serial monitor. Applications such as an autopilot for a drone are out of the question.

What you need to do is to process the raw data in order to extract useful information.

Processing raw motion sensor data can be done  in a variety of ways. 

At the bottom end of the complexity scale, you can use simple techniques such as smoothing or averaging the raw accelerometer values (also, see here). Such techniques can help clear up the noise and generate more reliable and useful motion information.

The most reliable motion information come by combining multiple types of motion sensors.

The MPU6050 contains an accelerometer and a gyroscope.

There are various algorithms that take input from both the accelerometer and the gyroscope on the sensor to calculate reliable and useful motion information. Because these algorithms take input from multiple sensor types, we often refer to them as "fusion" algorithms. 

You can then use the output of these algorithms to control a drone or or to navigate aircraft (see inertial navigation system).

This page contains an excellent tutorial on how to combine data from the gyro and accelerometer to derive useful values.

You may also want to read about the Kalman filter; this is an algorithm used widely for guidance, navigation, and control of all sorts of vehicles, from planes to spacecraft.

Here is an implementation of the Kalman filter for the Arduino.

Luckily, there are motion sensing devices that solve this problem by running fusion algorithms internally. This means that your Arduino will be able to receive clean and reliable "fused" motion information from the module, instead of noisy motion data.

The MPU-6050 is such a device.

It contains an integrated Digital Motion Processor (DMP); this is a module that does all sorts of motion-related calculations inside the chip, very efficiently and quickly. 

The MPU-6050 DMP is relatively sophisticated. It supports 3D motion processing and gesture recognition algorithms. It even reports temperature, and if you connect an external magnetometer, the DMP can also give you heading information. The external magnetometer can be connected to the MPU-60505 via an auxiliary I2C interface, so again you don't have to tie up the Arduino for this additional sensor.