AI-based Computer Vision to Enable Robotic Automation in High Mix Low Volume Assembly

Automating assembly processes in High-Mix, Low Volume (HMLV) manufacturing remains challenging, especially for Small and Medium-sized Enterprises (SMEs).
Consequently, many companies still rely on a significant amount of manual operations with an overall low degree of automation. The emergence of AI-based algorithms offers potential solutions, enabling assembly automation compatible with multiple products and maintaining overall production flexibility. However, the adoption of such technologies in the HMLV industry is low. There is currently no universal approach for effectively utilising the technologies in assembly automation in the context of HMLV products.
Furthermore, many approaches in existing research are non-industrial-oriented approaches, and they lack real-world implementation. This research presents a multidisciplinary approach to leverage learning-based Computer Vision (CV) methods to enable the automation of assembly processes in SMEs operating in an HMLV environment. With the proposed procedure, it is possible to identify process-relevant parameters critical for the automation of a given product-process combination. By developing and implementing learning-based CV models, these parameters are determined and made available for an automation system. Several experiments show the models' performance, flexibility, and robustness necessary for implementation in HMLV processes, demonstrating their suitability. The procedure is validated in three industrial use cases, showcasing its successful application. Following the procedure, several CV models are developed for a combined pick & place and glueing process and implemented on a technology demonstrator reaching Technology Readiness Level 4.
In the other use cases, the models developed with the procedure indicate high performance and mark the baseline for their process automation. The application in an industrial context can lead to increased productivity, higher quality, and reduced rework/scrap, securing the competitiveness of SMEs in a global market.