Most IMUs now a days have built in sensor fusion using kalman filters. For instance, Mbientlab-MetamotionR IMU. Now i would be getting angular rate from gyroscope. So how the sensor fusion works in this case (built in sensor fusion)? because if i want to get angle of rotation then i integrate the angular rate and get drift in the measurement.
Hello,
Indeed, more and more IMU embed fusion algorithms to determine attitude angles.
The sensor fusion manages to un-drift the integration of angular rates by making the assumption that, in average, the accelerometer measures the gravity, giving the downward direction (roll-pitch angles). This is true only if the sensor is not in constant acceleration (including centrifugal acceleration due to non-zero angular velocity) and this is usually the case with our robots. For the Yaw angle, some IMU also include a magnetic compass to unbiais it.
The IMU we use in the ODRI project includes an EKF algorithm outputting attitude directly.
Hoping this helps,
Best,
Thomas F.
1 Like