Sensor Fusion (MPU6050 + HMC5883L) || Kalman Filter || Measure Pitch, Roll, Yaw Accurately

𝗩𝗶𝗱𝗲𝗼 𝗗𝗲𝘀𝗰𝗿𝗶𝗽𝘁𝗶𝗼𝗻:
Discover how to accurately measure 3D orientation angles-Pitch, Roll, and Yaw-using the MPU6050 Accelerometer-Gyroscope and HMC5883L Magnetometer with an ESP32 microcontroller. In this detailed tutorial, we explore sensor fusion technology to combine data from these sensors, overcoming individual limitations like noise, drift, and interference. You’ll learn to implement a Kalman filter for enhanced accuracy and stability, calibrate the sensors for real-world use, and adjust yaw readings using your location’s magnetic declination.
This project includes step-by-step guidance on wiring the modules, writing code without using external libraries, and processing raw sensor data to calculate precise angles. The tutorial also demonstrates real-time visualization of Pitch, Roll, and Yaw values on an OLED display. Perfect for robotics, drones, and navigation systems that require precise orientation tracking. Full written tutorial and source code are linked in the description. Have questions? Let us know in the comments!
𝗪𝗿𝗶𝘁𝘁𝗲𝗻 𝗧𝘂𝘁𝗼𝗿𝗶𝗮𝗹 & 𝗚𝘂𝗶𝗱𝗲: how2electronic...
....................................................................................................................................................................................................................................
Drop a like if you liked this video.
Don't forget to subscribe to our channel for more Electronics projects and tutorials.
Website: how2electronic...
Facebook: / howtoelectronicsfb
Instagram: / howtoelectronics
Twitter: / how2electronics

[@LightSpeed_999]

Thank you for your great demonstration. I learned a lot from it. Great 👍

[@aknkaragoz9754]

Without overstepping, I would like to offer a few suggestions. First, set the filter window size. For example: #define WINDOW_SIZE 15 float pitchBuffer[WINDOW_SIZE] = {0}; float rollBuffer[WINDOW_SIZE] = {0}; float azimuthBuffer[WINDOW_SIZE] = {0}; int bufferIndex = 0; // It would be appropriate to keep the WINDOW_SIZE between 5 and 20. Then, add a moving average filter. For example: void applyMovingAverage(float &pitch, float &roll, float &azimuth) { pitchBuffer[bufferIndex] = pitch; rollBuffer[bufferIndex] = roll; azimuthBuffer[bufferIndex] = azimuth; bufferIndex = (bufferIndex + 1) % WINDOW_SIZE; pitch = 0; roll = 0; azimuth = 0; for (int i = 0; i < WINDOW_SIZE; i++) { pitch += pitchBuffer[i]; roll += rollBuffer[i]; azimuth += azimuthBuffer[i]; } pitch /= WINDOW_SIZE; roll /= WINDOW_SIZE; azimuth /= WINDOW_SIZE; } add loop: azimuth = atan2(correctedY, correctedX); azimuth += magneticDeclination; // Normalize azimuth to the 0-360 range if (azimuth < 0) azimuth += 2 * PI; if (azimuth > 2 * PI) azimuth -= 2 * PI; azimuth = azimuth * 180.0 / PI; // Convert to degrees // Apply moving average filter to pitch, roll, and azimuth applyMovingAverage(pitch, roll, azimuth); This way, you will be less affected by vibrations and quick direction changes. Please let me know the results if you try it. Best regards.

[@sulcusulnaris]

It's actually a small wonder that such things can be built into such a small component and that it is robust against vibrations, shocks etc.

[@1988trevor]

thank you. it's great for the small drone

[@torstenbehrendt870]

Great stuff. If I observed correctly I saw the the yaw changing when you only rolled the boards along its long axis. If my observation is correct, could you add the code lines to make this tilt correction, too?

[@reddawn87]

Wow this could not have come at a better time. I’m struggling with an MPU9250 for a project due on Friday. Can’t get the magnetometer to function properly to Measure the rotated angle. Thanks!

[@HowtoElectronics]

The MPU9250 has a similar structure, but the register addresses is different. I also had hard time in reading magnetometer reading from MPU9250. Tried multiple MPU9250, none of them.worked.

[@reddawn87]

@@HowtoElectronicsoh dear this does not bode well.. possible to do sensor fusion only using the 6050? I just need it to give a reference point for a digital compass.

[@reddawn87]

I am able to read the angle but it does not update in the void loop.

[@reddawn87]

I got my MPU9250 to work ultimately, ended up using an I2C scanner to find the IMU, then TwoWire to run it in parallel to my oled screen. I used hideakitai's examples to connect and calibrate it first, then managed to work. One thing I discovered, ESP32-S3 has an issue with pin reassignment in V3 of the Board Manager. I had to downgrade to 2.0.14 to get it to work.

[@HowtoElectronics]

Great man, you discovered so many new things. And I to find out you got it working. Congratulations

[@amrzakaria5290]

Good job , Thanks a lot.

[@HowtoElectronics]

Glad to hear it was helpful!

[@rehan_alammmblogs]

Nice ❤

[@aravindshreyas3571]

Hi, Thanks for the tutorial. I have few questions: 1. Will there be a gyro Drift after using KF? 2. Instead of two different boards, can the same be done using MPU 9250 ?

[@aravindshreyas3571]

I can use Arduino instead of ESP right?

[@HowtoElectronics]

There is improvement in gyro drift with time. You can use Arduino instead of esp as well. The MPU9250 has different register address and code is different for them.

[@aravindshreyas3571]

Thanks for the reply! Does this mean its impossible to eliminate gyro drift?

[@HowtoElectronics]

The gyro drift is already eliminated here. However there is always slight noise in mpu6050 due to vibration and some minor movements. A great alternative for MPU6050 is BMI160.

[@TobiKellner]

Hey, how complex would it be to integrate (RTK) GPS into this, for a fusion of position and orientation?

[@athulakulasinghe1957]

What is best for tvoc and eco2 ENS160 OR SGP30

[@sltechgalaxy1677]

very insightfull, make a tutorial to measure position using any imu sensor please

[@HowtoElectronics]

It's a good idea. I will make one soon.

[@sltechgalaxy1677]

@HowtoElectronics thank you dear

[@maheshmustapure7668]

Can you make video on esp32s3 with ov7670 on webserver streaming Or something related to that

[@danialothman]

Make sure pins are not magnetic or else it might disturb the magnetometer

[@cyber_karthik]

How to provision esp8266/32 using custom mobile phone

[@ΝΕΚΤΑΡΙΟΣΚΟΥΡΑΚΗΣ]

how much drift after 30 minutes?

[@HowtoElectronics]

Kalman Filter is used to normalize the reading and remove drift.

[@ΝΕΚΤΑΡΙΟΣΚΟΥΡΑΚΗΣ]

@HowtoElectronics I will try, again your way, thank you!

[@math8149]

Nowadays it's hard to get hmc5883.... So please make a video about fusing the sensor of mpu 6050 and qmc5883 sensor... Asap

[@HowtoElectronics]

QMC5883 and HMC5883 both are same. You can work with any of these....

[@math8149]

@HowtoElectronics you are partially correct.... Register addresses and set mode values aren't the Same..... And Thank you so so much... I really needed this video... For a long time, I was expecting someone to use the Kalman filter to fuse the accelerometer gyroscope and magnetometer sensor data... I couldn't see anyone doing it... So thank you again... And a small request (I know it's not small), If you can make code for an extended kalman filter to Fuse the mpu 6050 data and qmc 5883 data... It would be very useful to me and others as well. I was trying to make a model for EKF for a long time ... But I ended up failing...

[@HowtoElectronics]

Thank you for your detailed feedback and kind words! Regarding the Extended Kalman Filter (EKF) to fuse MPU6050 and QMC5883 data, I understand its importance. I will definitely give it a try, and if I can successfully implement it, I will post a video explaining the process step-by-step. Stay tuned, and I appreciate your patience!

[@rokasbarasa1]

Looks very innacurate based on the data