0

Currently, we are using BNO055 in one of our projects. The IMU is placed next to the dc motor due to space constraints within the hardware setup. Due to motors vibrations, we are applying a low pass filter on quaternion values read from this (https://github.com/adafruit/Adafruit_BN ... awdata.ino) script. We have set 5 Hz as a cut-off frequency of the filter. We have also placed IMU on Sorbothane (damping material) to minimize the vibrations. However, we are still selling the error in the orientation.

What could be done to reduce the impact of motor vibrations on IMU both from a software and hardware point of view? Any inputs are highly appreciated.

Veeresh
  • 1

1 Answers1

1

Motor vibration may not be the only problem here. Orientation estimation can go wrong due to multiple factors like,

  1. Bias due to incorrect calibration. Keep the sensor level and idle. Make sure Gyroscope reads close to (0,0,0) on average. Accelerometer should read either (0,0,9.81) or (0 0 -9.81) m/s^2 depending on the convention.
  2. Bias due to temperature changes. Even a 10 deg change in PCB temperature can change the bias in Gyro by 0.3 dps (according to the datasheet)
  3. Motor noise. Seems like you have already tried reducing this one.

If none of them work, you could try implementing your own Kalman or complementary filter based on the raw data from Gyro, Accel and mag. This way you can be sure about calibration process, estimator gains, how the estimator works.

If implementing Kalman filter is difficult, you could try this AHRS filter block/algorithm given here, https://in.mathworks.com/help/nav/ref/ahrsfilter-system-object.html

Manoj
  • 86
  • 5