MATLAB Answers

Calculate Measurement Noise for Sensor Fusion for Inertial Navigation ["Pose Estimation From Asynchronous Sensors", "IMU and GPS Fusion for Inertial Navigation"]

7 views (last 30 days)
Christoph Müßig
Christoph Müßig on 2 Nov 2019
I'm following the two guides "Pose Estimation From Asynchronous Sensors" [] and "IMU and GPS Fusion for Inertial Navigation" [], since I want to use self recorded data for Inertial Navigation at a highschool project.
In both examples values for measurement noise and process noise are used, without really explaining how they were found. One example just states the measurement noise depends on the sensor accuracy, the other states that they can be found in the datasheet of the sensors. And with the process noise, the only hint given is from the second example that they "are determined empirically using parameter sweeping to jointly optimize position and orientation estimates from the filter".
My first question is if I interpret the statement for the process noise right, that I just have to try different values and compare the results? And if so, is their a better solution than just enter values manually?
And with the measurement noise I searched everywhere in all the datasheets of my sensors, but I wasn't able to find these values. In which unit are the values or do they even have a unit? I haven't found it anywhere in the documentation. Or is there a calculation - I wasn't able to find any useful information about such a formula in the internet - based on sensor stats like measurement range, resolution, noise density and temperaturebias, which I specified earlier in the programm?
Thanks a lot for helping a beginner in Sensor Fusion and the using of Kalman filters.

Answers (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!