Geometric Modeling and Artificial Intelligence for Design and Manufacturing Automation

本项目主要研究三维空间推理及其在智能工程系统中的应用。集成几何建模和人工智能(AI)领域的理论和技术，以解决机电产品生命周期中设计、制造等活动中出现的难题。这项研究的主体是先前国家科学基金资助的关于可加工特征的自动识别和利用坐标测量机(CMM)检测机械零件的可达性分析的自然续篇。南加州大学的特征识别器将集成在下游，与实验工艺规划器和用于驱动机床的商用NC(数控)刀具轨迹生成器集成在一起。它还将与上游的基于特征的设计系统集成在一起。现有的特征搜索器非常有前途，因为它可以处理特征之间复杂的体积相互作用，并产生满足严格的可加工性条件的去除体积。它将被重新设计，以便以增量方式运行，从而向设计师提供即时的制造反馈。它还将被扩展，以提取工艺规划者和NC代码生成器所需的几何信息。对检验计划的研究将寻求在给定机械部件的公差实体模型的情况下，全自动地产生用于驱动坐标测量机(以及潜在的其他设备，如激光探头)的完整程序。尺寸检验与公差测量密切相关，并将建立在南加州大学正在进行的公差规范语义工作的基础上。还将讨论测量解释(软测量)的问题。

This project is centered on three-dimensional spatial reasoning and its applications in intelligent engineering systems. Theories and techniques from the fields of geometric modeling and artificial intelligence (AI) are integrated so as to attack difficult problems that arise in design, manufacturing and other activities in the life cycle of mechanical and electromechanical products. The main body of the research is a natural sequel to previous, National Science Foundation supported work on automatic recognition of machinable features, and on accessibility analysis for the inspection of mechanical parts with Coordinate Measuring Machines (CMMs). USC's feature recognizer will be integrated downstream, with experimental process planners and with commercial NC (Numerical Control) cutter-path generators for driving machine tools. It will also be integrated upstream, with feature-based design systems. The existing feature finder is very promising because it can handle complex volumetric interactions between features, and produces removal volumes that satisfy stringent conditions for machinability. It will be redesigned so as to operate incrementally, thereby providing immediate manufacturing feedback to a designer. It will also be extended so as to extract the geometric information required by process planners and NC code generators. Research on inspection planning will seek to produce, fully automatically, complete programs for driving CMMs (and potentially other equipment such as laser probes), given toleranced solid models of mechanical parts. Dimensional inspection is closely related to tolerancing, and will build upon on-going work at USC on the semantics of tolerance specifications. Issues of measurement interpretation (soft gaging) also will be addressed.