"视觉惯性传感器时间同步标定方法及其在SLAM技术中的应用"

Simultaneous localization and mapping (SLAM) technology based on visual-inertial information fusion has gained significant attention in the field of robot navigation. The synchronization of multimodal data is crucial for data fusion algorithms, and the key to achieving this is accurately obtaining the timestamps of data collected by different sensors. While the timestamp of visual sensors can be accurately obtained through hardware methods, the timestamp of inertial sensors is usually replaced by output time, leading to a decrease in the precision of visual-inertial fusion algorithms. To address this issue, a method for synchronizing the timestamps of visual and inertial sensors is proposed. Firstly, a calibration device based on a planar pendulum is designed, allowing the visual and inertial sensors to independently capture data during planar pendulum motion, and time stamps are added based on the same clock. Secondly, a method based on the least squares estimation of the pendulum's center of mass angular displacement function and angular velocity function is proposed. Lastly, by comparing the phase difference of the two functions, the output delay and timestamp of the inertial sensor can be obtained. Experimental results demonstrate that the proposed method has a repeatable calibration mean square error of 0.018 ms, validating its effectiveness. In conclusion, the proposed method provides a solution for synchronizing the timestamps of visual and inertial sensors, thus improving the precision of visual-inertial fusion algorithms. This has implications for the development of SLAM technology in the field of robot navigation.