在MLA曝光工艺中，曝光点的数量庞大，通过高倍率显微镜配合人工目检来判定曝光质量耗时耗力，造成工艺成本偏高。为了解决这个问题，设计了一种便于检测的圆环形图案并引入深度学习中的目标检测Yolov5模型，一定程度上能够取代人工目检，完成对曝光质量的快速判定。基于上述方法，分析了不同光刻胶厚度之下，线能量密度的最优区间与光刻胶的剖面倾角。并在同等线能量密度下通过圆度判定曝光图案失真情况。在本研究的MLA曝光工艺中，选取光刻胶厚度、激光曝光功率以及加工平台移动速度作为自变量，评价曝光合格率、光刻胶剖面倾角以及曝光圆度等加工质量参数具有重要的工程意义。

Overview: Driven by the rapid development of national optical projects such as laser nuclear fusion and aerospace telescopes, as well as high-end civilian fields such as advanced instruments and optical lenses, the requirements for full-frequency domain processing errors and surfaces of optical components are becoming more and more stringent. At this stage, the optical components generally need to go through rough grinding, fine grinding, polishing and coating, and other processes, and their surface quality mainly depends on the defect removal ability and error control level of the polishing process. Whether the fine grinding process can obtain better surface shape accuracy and low surface/subsurface damage suppression determines the processing efficiency, and the ultra-precision processing manufacturing equipment is the premise of the realization of ultra-precision machining of the optical components. So far, all countries in the world have invested in the research and development of optical ultra-precision grinding and polishing technology, and have developed more relatively mature high-precision grinding and polishing equipment, which can better meet the processing needs of most of the current optical components. For the core equipment and key technologies required for ultra-precision manufacturing, China has long relied on imports. In order to break through the bottleneck restricting the development of ultra-precision technology in China at this stage, under the traction and drive of the national large-scale engineering project, China has made remarkable progress in optical ultra-precision manufacturing equipment and technology. However, for the optical ultra-precision technology and equipment, there is still a certain gap between China and the international advanced level, and it is necessary to continue to strengthen the research. In addition to the high-end grinding and polishing equipment necessary for the ultra-precision machining of optical components, it is also necessary to strengthen the technical level of a series of key supporting units, such as ultra-precision grinding and polishing processing technology, high-end key functional components, intelligent monitoring technology of processing environment, efficient ultra-precision machining tools, processing and inspection path planning and compensation processing strategies, computer-aided manufacturing and testing software, etc. The research, development, and application of these technologies are related to the development of high-end manufacturing in the civilian fields and national defense fields, and are also the focus of the country. This paper mainly focuses on the ultra-precision machining of large-diameter optical aspherical components. Starting from the grinding and polishing process route, this paper introduces the long-term research progress of the Precision Engineering Laboratory of Xiamen University in the field of large-diameter optical aspherical component processing, and introduces in detail the technical and system achievements such as ultra-precision grinding and polishing equipment, robot-assisted grinding and polishing, equipment intelligent monitoring system, processing technology and control software.

针对六自由度串联式关节机器人气囊抛光系统因刚度不足引起的加工振动以及引入中频误差的问题，以IRB 6700机器人作为研究对象，基于Ansys Workbench建立模态分析模型，并结合实验分析机器人气囊抛光系统工况频带内动态特性，实验与仿真结果共同表明，机器人气囊抛光系统在工况频带至少存在5阶模态，且共振时机器人末端抖动幅值为mm级，机器人加工严重受限。同时针对机器人气囊抛光系统先进光学元件抛光工艺应用，设计一种阻尼抑振气囊工具头，与普通气囊工具头进行定点抛光与整面抛光对比实验。结果表明：抑振气囊头定点抛光斑粗糙度与频谱幅值普遍低于普通气囊工具头，引入的中频误差较一般气囊工具头低40%，抛光优化效果显著。

基于气囊抛光技术和工业机器人平台开发光学元件精密抛光系统，既能满足光学元件快速抛光环节的高效率和高精度的要求，又可以降低开发成本，是极具潜力的抛光设备开发方案。气囊抛光具有稳定的且确定的材料去除特性，通常要求抛光斑稳定性在90%左右。针对机器人气囊抛光系统在多步离散进动抛光过程中机器人末端刚度对气囊抛光稳定性的影响展开研究，通过建立机器人末端刚度矩阵，获得机器人末端变形；基于Preston理论，建立含变形误差的气囊抛光去除函数。最后设计4步离散定点抛光实验验证机器人气囊抛光系统稳定性。根据结果可知抛光斑在XY截面轮廓线上皆呈类高斯形状，且XY截面轮廓线基本一致，具有比较好的重合度；对比不同抛光位置的截面轮廓线，其相对误差小于5%，由此可验证机器人气囊抛光系统在离散进动抛光时具有较好的稳定性。