基于共面4点的机器人位姿求解及其不确定性分析

Pose estimation of robots based on coplanar four points and its uncertainty analysis

针对工业机器人视觉伺服控制系统的位置定位问题,提出了一种采用共面4点的单目机器人视觉系统的摄像机位姿求解方法。为评价这种单目机器人视觉系统的定位精度,对定位不确定性的评定方法进行了研究,并在深入分析机器人视觉系统平面定位精度不确定性的基础上,提出将极大似然抽样一致性与上述位姿求解方法相结合,以改善机器人的定位精度。实验证明,该方法在抗噪声能力、对外点稳定性方面优于传统的最小二乘法,有利于提高视觉系统的定位精度和鲁棒性。

Considering positioning problems of industrial robot vision servosystem,a method based on coplanar 4 points pattern for camera pose estimation of monocular vision robots was presented.To evaluate the positioning accuracy of this robot vision system,the evaluation method based on positioning uncertainty was studied.And by analyzing uncertainty of planar positioning accuracy of robot vision system,a method combining maximum likelihood estimation sample consensus and the above pose estimation algorithm was presented.Experimental results showed that the proposed method was superior to the traditional least square algorithm on anti-noise and resisting out point ability.It could effectively enhance the robustness and accuracy of the robot vision system.