Abstract: An automatic assembly system was developed aiming at solving the problems of weak automation and poor assembly consistency of miniature thread pairs during gimbal assembly. The extraction method of the central positions for the threaded parts and the tightening strategy of multiple miniature thread pairs were studied. Firstly， an improved clarity evaluating function was proposed， and the auto focus of machine vision device was realized based on the evaluating function. Then， aiming at the asymmetry of the end surface features of the threaded shafts and the threaded holes of the inner frame， the variable focused and regional image acquisition method was proposed for extracting the center of the inner and outer threaded parts. Finally， combining the relationship between the input rotation angle and the axial displacement of the inner and outer frames in the loading and unloading processes， the tightening strategy called cross unloading-loading in group was proposed for multiple threaded joints. The system prototype for assembly experiments was built， and the experiments were carried out. The results show that the coaxiality between outer frame and base frame is less than 5 μm， the axial displacements of the inner frame and the outer frame are less than 3 μm， which may meet the coaxiality requirements of the gimbal. 

Key words: gimbal, miniature screw thread pair, automatic assembly, machine vision, tightening strategy

摘要： 针对万向支架微小螺纹副装配自动化程度低、一致性差的问题，研制了一套自动装配系统，并对装配过程中的微小螺纹零件中心位置提取方法和多个微小螺纹副之间的装调策略进行了研究。首先，提出了一种改进的清晰度评价函数，并基于该函数实现了机器视觉装置的自动聚焦。然后，针对螺轴以及内环螺纹孔的端面特征不对称问题，提出了变聚焦位置分区域图像采集的内外螺纹中心提取方法。最后，结合加载和卸载过程中输入转角和内外环轴向位移量之间的关系，制定了螺轴分组交叉卸载-加载的装调策略。搭建系统样机进行装配实验，结果表明：外环支架偏心量小于5 μm，内环及外环的轴向窜动量均小于3 μm，满足万向支架的同轴度等性能要求。

关键词: 万向支架, 微小螺纹副, 自动装配, 机器视觉, 拧紧策略