一种视觉惯性系统位姿高精度一体化标定方法

A High-Precision Integrated Calibration Method for Position and Attitude of Visual-Inertial System

在视觉惯性定位系统中,传感器位姿关系的标定对于实现精确空间定位至关重要,针对现有标定方法对多传感器系统缺乏集成性、标定精度受限等问题,提出了一种视觉惯性系统位姿高精度一体化标定方法。通过精密三轴转台提供角度基准,基于重力矢量不变性和匀速圆周运动下向心加速度数值的一致性求解惯性测量单元(IMU)与转台之间的外参,利用转台构建控制场为相机标定提供空间角度约束,联合优化求解无重叠视场多相机内外参。仿真和实验结果表明,该方法具有较高的标定精度和稳定性,在多相机IMU系统组合定位测试中,与经典标定方法Kalibr相比,本文方法系统运动轨迹拟合轴线的角度偏差降低40.32%,距离偏差降低18.93%,可满足高精度视觉惯性定位系统的标定需求。

In the visualinertial positioning system,the calibration of the sensor pose relationship plays a crucial role in realizing accurate spatial positioning.Existing calibration methods lack integration for multisensor systems,and the calibration accuracy is limited.In this paper,a highprecision integrated calibration method for the position and attitude of the visualinertial system is proposed.A precision threeaxis turntable is used to provide the angle reference.The extrinsic parameters between the inertial measurement unit(IMU)and turntable are solved based on the invariance of the gravity vector and the consistency of centripetal acceleration values.The control field is constructed by the turntable to provide spatial angle constraints for camera calibration,and the intrinsic and extrinsic parameters of nonoverlapping cameras are jointly optimized.Simulation and experimental results show that this method has high calibration accuracy and stability.In the combined positioning test of the multicamera IMU system,compared with the classical calibration method Kalibr,using the calibration results of this method,the angle deviation of the system motion trajectory fitting axis is decreased by 40.32%and the distance deviation is decreased by 18.93%,which can meet the calibration requirements of highprecision visualinertial positioning systems.