I am doing an EKF (Extended Kalman Filter) to fusion Visual Odometry with IMU information. I am using a ROS package called ethzasl-msf (I am using the multi sensor fusion because then I have to add some GPS data).

But the problem is that I need the noise and bias of the accelerometer and gyroscope. I am using an AR.Drone 2.0, and I don't have the IMU model, so I can't even guess those values.

As specified in the .yaml document, those values for an ADIS 16448 are:

  • accelerometer noise: 0.0022563
  • gyroscope noise: 0.04
  • accelerometer noise bias: 8e-5 (it is not fixed)
  • gyroscope noise bias: 3e-6 (it is not fixed)

I suppose that the noise of AR.Drone 2.0's IMU is worse, because it's a cheap quacopter. But I couldn't find any information about them. What values do you recommend me to use?


  • $\begingroup$ Generally one would get the device and measure them...it's the only surefire way to get an answer...otherwise I would just go with the data you have and model with that, random values you find, or are given are no better or worse than what you already have. And there are a lot of very good 'cheap' IMUs...hard to know what your drone uses. $\endgroup$ Aug 7, 2019 at 17:29
  • $\begingroup$ But do you have a range or something of common noise values? I don’t know if I have to change the noise 10 times, 100 times or 0.01 times $\endgroup$
    – Unnamed
    Aug 9, 2019 at 16:38

1 Answer 1


Bias should be automatically measured while the sensor is stationary at the beginning of each run. This is simply because the biases have different starting values at each run according to the ambient temperature and chip temperature. But I am not sure why they put bias values in the yaml file. You need to have a look at the code to see if they are really using the input bias.

accelerometer noise, gyroscope noise are just weighting factor. You might able to run the code with your new IMU with the values from ADIS 16448. If it does not work, the simplest way is running a calibration between IMU and camera which will give you those values. You can use kalibr which is from the same lab.

  • $\begingroup$ How can I measure the bias? $\endgroup$
    – Unnamed
    Aug 9, 2019 at 16:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.