关键词: 工业机器人, 视觉系统, 位姿求解, 不确定性, P4P问题, 极大似然抽样一致性

Abstract: 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.

Key words: industrial robots, vision system, pose estimation, uncertainty, P4P problem, maximum likelihood estimation sample consensus