Knowledge Graphs in Manufacturing Robots: Task Execution and Knowledge Maintenance

When manufacturing robots perform assembly, handling, and other tasks, they need to comprehensively utilize product structures, process constraints, equipment capabilities, and on-site states. Knowledge graphs can organize these multi-source heterogeneous information as entities and relationships, providing a knowledge foundation for object matching, operational parameter inference, skill invocation, and task planning. Existing research has applied graph knowledge to grasping parameter selection, assembly task decomposition, and program generation, reporting execution results in some physical robot tasks. The role of graphs is closely related to the accuracy of object information, skill definitions, and equipment configurations, and their execution results need to be analyzed in conjunction with corresponding experimental conditions. This paper reviews the application of knowledge graphs in manufacturing robot tasks, systematizes the sources and organization methods of relevant knowledge, focuses on analyzing their role in workpiece selection and equipment allocation, operational parameters and skill selection, task planning and program generation, and further discusses knowledge maintenance and feedback utilization during task execution.