fusing raw IMU data, specifically angular velocities, and linear accelerations. The data was first read from a CSV file from a rosbag collected from bagging different topics from a moving drone indoors and within VICON motion capture system, and the time, angular velocities, and linear accelerations were extracted. A sampling frequency of 85 Hz was chosen, and the noise estimation parameters of the imufilter were set to 0.001 for GyroscopeNoise and 0.01 for AccelerometerNoise. The IMU data was then processed through the imufilter to obtain the quaternion estimates, which were then converted to Euler angles. Finally, the fused Euler angle estimates were plotted and compared against the mavros orientation estimates.
They align very well; however, the output shows tiny drift when compared to the MAVROS data. This
drift might be due to inaccuracies in the IMU data, noise, or limitations of the filtering method. Further
tuning of the noise estimation parameters or the use of additional sensor data could help in reducing the
drift and improving the accuracy of the fused orientation estimates.
fusing raw IMU data, specifically angular velocities, and linear accelerations. The data was first read from a CSV file from a rosbag collected from bagging different topics from a moving drone indoors and within VICON motion capture system, and the time, angular velocities, and linear accelerations were extracted. A sampling frequency of 85 Hz was chosen, and the noise estimation parameters of the imufilter were set to 0.001 for GyroscopeNoise and 0.01 for AccelerometerNoise. The IMU data was then processed through the imufilter to obtain the quaternion estimates, which were then converted to Euler angles. Finally, the fused Euler angle estimates were plotted and compared against the mavros orientation estimates. They align very well; however, the output shows tiny drift when compared to the MAVROS data. This drift might be due to inaccuracies in the IMU data, noise, or limitations of the filtering method. Further tuning of the noise estimation parameters or the use of additional sensor data could help in reducing the drift and improving the accuracy of the fused orientation estimates.