基于互相关和旋转约束的视觉惯性里程计在线时间校准算法  

Online temporal calibration based on cross-correlation androtation constraints for visual-inertial odometry

作　　者：蒙军杰 熊军林 Meng Junjie;Xiong Junlin(School of Information Science&Technology,University of Science&Technology of China,Hefei 230026,China)

机构地区：[1]中国科学技术大学信息科学技术学院,合肥230026

出　　处：《计算机应用研究》2025年第1期288-292,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(62273320)。

摘　　要：在融合相机和惯性测量单元(IMU)的数据推测机器人的运动轨迹时,传感器测量记录的时间点对用于估计轨迹的视觉惯性里程计(VIO)的鲁棒性和准确性至关重要。然而,由于传感器数据到达接收端的延迟存在差异,图像数据流和IMU数据流之间通常存在不可避免的时间偏置,为此提出了一种基于互相关和旋转对齐的视觉惯性里程计在线时间校准的算法。首先使用对极几何和预积分算法分别得到相机和IMU各自的相对位姿,并计算出相机的角速度;然后根据相机与IMU的角速度进行互相关计算,得到初步的时间偏置估计;最后利用相机和IMU相对位姿进行旋转约束,通过优化误差函数得到更精确的相对时间偏置估计,该时间偏置值随后用于平移传感器的时间轴以进行校准。实验表明,该算法能够减缓时间偏置对里程计精度带来的影响,并使得VIO能够在具有更大时间偏置范围的数据流下稳定运行。In the fusion of camera and inertial measurement unit(IMU)data for inferring the robot’s trajectory,the timing of sensor measurements plays a crucial role in the robustness and accuracy of the visual-inertial odometry(VIO).However,due to differences in the delay of sensor data reaching the receiver,an unavoidable time offset exists between the image data stream and the IMU data stream.To address this issue,this paper proposed an algorithm based on cross-correlation and rotation constraints for online temporal calibration of VIO.Firstly,it used the epipolar geometry and preintegration algorithms to obtain the relative poses for both the camera and the IMU.Then,it evaluated cross-correlation based on the angular velocities to obtain an initial estimate of the time offset.Finally,it utilized the relative pose between the camera and IMU to impose rotation constraints and optimize an error function,leading to a more accurate estimation of the time offset.It employed this time offset va-lue to calibrate the sensors.Experimental results show that the proposed algorithm effectively mitigates the impact of time offset on the accuracy of odometry trajectories,enabling the system to reliably operate when significant time offsets exist in the data streams.

关 键 词：在线时间校准 旋转约束 视觉惯性里程计 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]