单目相机-IMU-机械臂的空间联合标定方法

A space joint calibration method of monocular camera-IMU-manipulator

将单目相机和惯性测量单元(IMU)固定在机械臂末端,可以有效地协助机械臂进行运动估计和场景感知。单目相机、IMU和机械臂之间可靠、准确的标定是实现随动感知的前提。但是,传统的标定方法存在耦合误差大、标定一致性差和6自由度激励不充分的问题。为此,本文提出了一种新颖的单目相机-IMU-机械臂的空间联合标定模型,该模型能够将旋转分量的误差和平移分量的误差完全解耦。此外,创新性地设计了一种螺旋式运动轨迹来标定感知模块,采用这种运动轨迹能够避免6自由度激励不充分和一致性差等问题。在实验环节,本研究搭建了单目相机-IMU-机械臂空间联合标定平台,计算得到三轴姿态角和平移分量的均方根误差分别小于0.7°和7 mm,证明了所提标定方法的有效性。

Fixing the monocular camera and the inertial measurement unit(IMU)on the end of the robotic arm is an effective method to realize the motion estimation and scene perception.Reliable and accurate calibration between monocular camera,IMU and robotic arm is the prerequisite for realizing follow-up perception.However,traditional calibration method has the problems of large coupling error,poor calibration consistency and insufficient 6-DOF excitation.Therefore,this paper proposes a novel monocular camera-IMU-manipulator space joint calibration method,which can completely decouple the errors of the rotation component and the translation component.At the same time,this paper innovatively designs a spiral motion trajectory to calibrate the perception module.Using such a calibration trajectory can avoid the problems of insufficient 6-DOF excitation and poor consistency in traditional calibration methods.In the experimental link,a monocular camera-IMU-manipulator space joint calibration platform is built,and the root mean square errors of yaw-pitch-roll rotation and x-y-z translation are calculated to be less than 0.7°and 7 mm,respectively,which proves the effectiveness of the calibration method proposed in this paper.