An IMU/Visual Odometry Integrated Navigation Method Based on Measurement Model optimization

In order to improve the accuracy and reliability of the positioning and navigation system of the intelligent-motion objects in complex environment, and to solve the precision reduction problem of the integrated navigation system caused by the error accumulation of the visual odometry (VO) with motion and the inaccurate modeling of VO noise, an IMU/visual odometry integrated navigation method based on measurement model optimization is proposed. With Kalman filter as the framework, we constructs the state equation by the error recurrence model of strapdown inertial navigation system (SINS), and based on the characteristics of the relative pose output of VO, we consider VO as an angular velocity, line velocity and position sensor to construct the measurement equation. The method has been evaluated by KITTI odometry benchmark and the results indicate that comparing to traditional integration of position and attitudes, the measurement model of this method is more consistent with the noise characteristics of VO and thus can improve the accuracy and reliability of the integrated navigation system.