论文部分内容阅读
在分析移动机器人车载摄像机位姿的特殊性质的基础上,根据摄像机的等效转轴构造辅助旋转矩阵,利用该旋转矩阵将原始待分解本质矩阵和单应矩阵转换为一类简单的、可通过初等数学运算进行分解的本质矩阵和单应矩阵。仿真实验的结果表明,该车载摄像机位姿估计算法较传统方法具有更高的精度和更快的运算速度,对摄像机等效转轴的扰动也具有很好的鲁棒性。此外,分解出的可能解的数目较传统算法减少了一半,且在除诱导单应阵的空间景物平面与地面垂直的情况下,均能直接得到移动机器人的唯一转角,为移动机器人姿态控制提供了极大的便利。
Based on the analysis of the special properties of the pose of the mobile robot on-board camera, the auxiliary rotation matrix is constructed according to the equivalent axis of the camera. By using the rotation matrix, the original matrix and homography matrix to be decomposed are transformed into a simple class, Essential matrices and homography matrices for decomposition of mathematical operations. Simulation results show that the pose estimation algorithm of vehicle camera has higher precision and faster computing speed than the traditional method, and also has good robustness to the perturbation of the camera equivalent shaft. In addition, the number of possible solutions is reduced by half compared with the traditional algorithm, and the only rotation angle of the mobile robot can be directly obtained except the plane of the singular space plane is induced to be perpendicular to the ground, providing the attitude control of the mobile robot Great convenience.