A hierarchical retrieval approach for automatically generating assembly instructions

Assembly instructions are critical for prompt fulfilment of assembly tasks. However, the design of such instructions is time-consuming and requires experience. Retrieve and reuse of previous cases shortens the design time and reduces design mistakes, whereas the traditional retrieval method encounters a bottleneck during encoding because the technical instructions include both structured and unstructured data. In this paper, we propose a hierarchical retrieval approach for automatic generation of assembly instructions based on previously used technical instruction cards. First, a case-based reasoning (CBR) method is employed to encode the assembly process and retrieve the suitable cases. Then, an improved weighted latent Dirichlet allocation text mining technique is applied to explore unstructured text topics and recommend the most optimal case. Finally, we utilize the proposed method to an automotive assembly process using data in 12,034 used instruction cards. The results demonstrate that technical instructions can be generated automatically for a specific topic using the proposed retrieval method. Compared to the traditional CBR method, the proposed hierarchical retrieval approach significantly improves the quality of new assembly instructions and the speed of generation.