I have a small quadcopter equipped with the MPU6050, which has a built-in accelerometer and gyroscope. It is able to sample the acceleration at 1 kHz and the angular velocities at 8 kHz. However, I only sample them at 200 Hz. I placed my drone indoors before a large fan to see how it responds, and the typical readings look like this:

enter image description here

These are raw signals. They have only been converted to the proper units. Why is the angular velocity so much smoother than the acceleration? The only guess I have is that the MPU6050 is averaging the measurements each time I sample: so for acceleration it can average 5 measurements (1000/200), while for the angular velocity it can average 40 measurements (8000/200).

  • $\begingroup$ What do the signals look like if you sample at other rates? It sounds like you think the sensor is doing some averaging, so the data should look noisier at higher sampling rates. $\endgroup$ – Robin Bennett Nov 17 '20 at 9:12
  • $\begingroup$ I haven't investigated in depth, but there is no mention of average or mean in the datasheet, so I think it's unlikely that this is happening. $\endgroup$ – ifconfig Nov 17 '20 at 18:41
  • $\begingroup$ Were the motors powered during the logging event in from of the fan? $\endgroup$ – ifconfig Nov 17 '20 at 20:12
  • $\begingroup$ @ifconfig yes, the drone was flying. So I am expecting the accelerometer to be noisy, but I was also expecting the gyro to be noisy... $\endgroup$ – Thomas Wagenaar Nov 18 '20 at 8:01
  • $\begingroup$ @RobinBennett unfortunately the software does not allow me go above 200 Hz while flying. But good idea, I will try to do this with the drone simply in my hand $\endgroup$ – Thomas Wagenaar Nov 18 '20 at 8:01

A large change in acceleration doesn't necessarily mean a big change in velocity. Since the chance in velocity is the integral of the acceleration, the velocity will barely change if the acceleration spikes quickly for example.

Something that I don't quite understand is why the two graphs doesn't match. By definition, the acceleration should be negative when the velocity decreases, and positive when it increases. It doesn't look like that is the case here, even if the zero value on the y-axis is wrong. So it is possible that One of the sensors gives you the wrong readout.

  • 1
    $\begingroup$ Thanks for your answer! But the angular and spatial positions are independent so they don’t correlate as you propose, i.e. acceleration and angular velocity aren’t each others derivatives $\endgroup$ – Thomas Wagenaar Dec 19 '20 at 10:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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