Impact of the Earth rotation compensation on visual inertial SLAM

Aiming at the problem that the traditional IMU pre-integration model in the visual inertial SLAM system cannot be effectively adapted to the measurement accuracy of the inertial measurement unit (IMU) due to the improvement of its accuracy, resulting in the loss of the accuracy of the SLAM system in the construction of maps and navigation, this paper proposes a kind of IMU pre-integration model considering the compensation of the earth's rotation. The method derives a pre-integrated IMU model that includes Earth motion compensation by analyzing the Earth motion model, fusing the Earth motion model into the IMU position measurement model, and analyzing the IMU measurement noise and bias during motion. The validation results with different difficulty data sets show that the IMU pre-integration model after the introduction of earth motion compensation is better than the existing traditional pre-integration model in terms of robustness and accuracy, and its localization accuracy has been improved by an average of 40%, and the trajectory error has been reduced by an average of 45%.