汕头大学学报（自然科学版）

6D位姿是目标物体基于平面矢量和旋转矢量的位置和姿态.基于单目RGB相机进行6D位姿估计的传统方法由于标记点少、识别率较低等原因，易导致测量精度不高.为提高物体位姿测量的精度，论文提出了一种基于特征标记的物体6D位姿测量方法.首先，设计一种能提供更多标记点且识别率高的特征标记，并通过相机识别被测物体上的特征标记获得3D-2D点对.其次，基于相机成像原理设计物体6D位姿测量系统与方法.然后，建立求解工件位姿变化的旋转矩阵和位移矩阵模型.最后，基于所建立的旋转矩阵模型与欧拉角之间的关系求解工件6D位姿.实验结果表明，在0-170 mm位移范围内，沿X、Y、Z轴位移测量误差小于0.5 mm；在0-30°旋转范围内，其绕X、Y轴旋转角度测量误差小于0.8°；在0-45°旋转范围内，其绕Z轴旋转角度测量误差小于0.8°；测量物体绕X、Y轴旋转角度误差耗时小于1.5 s，绕Z轴旋转角度误差耗时小于2 s；测量物体沿X、Z轴位移误差耗时小于2 s，沿Z轴位移误差耗时小于2.5 s；能够满足实际生产要求.实验结果检验了论文所提方法的有效性，可用于实际生产环境中.

The 6D pose of an object is its position and attitude based on planar vectors and rotation vectors. Traditional methods for estimating the 6D pose of an object using a monocular RGB camera often result in low measurement accuracy due to reasons such as a lack of markers and low recognition rates. To improve the accuracy of object pose measurement, a feature marker-based method for measuring the 6D pose of an object is proposed. First, a feature marker that can provide more markers and has a high recognition rate is designed. 3D-2D point pairs are obtained by recognizing the feature markers on the object to be measured with the camera.Second, a method for measuring the 6D pose of an object is designed based on the principle of camera imaging. Then, models of rotation matrices and displacement matrices for solving the pose changes of the workpiece are established. Finally, the 6D pose of the workpiece is solved based on the relationship between the established rotation matrix model and Euler angles.Experimental results show that within a displacement range of 0-170 mm, the measurement errors of displacement along the X, Y, and Z axes are less than 0.5 mm; within a rotation range of 0-30°, the measurement errors of rotation angles around the X and Y axes are less than 0.8°;within a rotation range of 0-45°, the measurement error of rotation angle around the Z axis is less than 0.8°; the time consumed for measuring the rotation angle error of the object around the X and Y axes is less than 1.5 s, and the time consumed for measuring the rotation angle error around the Z axis is less than 2 s; the time consumed for measuring the displacement error of the object along the X and Z axes is less than 2 s, and the time consumed for measuring the displacement error along the Z axis is less than 2.5 s; it can meet the actual production requirements. The experimental results verify the effectiveness of the proposed method, which can be used in actual production environments.