Correlating machine vision and learning with robot handling in increasing productivity of airbags manufacturing

The automotive industry is permanently looking for increased productivity, improved flexibility, and customer satisfaction. Considering that the manufacturing of airbags is based mostly on textiles, there is a lot of manual work in the airbags process assembly; this is a major issue for productivity. The aim of the present, on-going work has been to study the benefits and limitations of utilizing machine learning (ML), video cameras, and robotic handling in order to reduce (or even eliminate) manual work in what concerns manufacturing processes of airbags. The approach has been based on using automation of processes in airbags manufacturing. In the initial stage of the in-development industrial process, the speed of the considered airbag manufacturing line was very close to a manual line. The paper briefly presents the optimization of the algorithms (using AI), the improvement of the process routes, as well as installing cameras to guide the robots to reach a better repeatability. This on-going procedure is briefly described. The final achieved results have included: 75% higher speed, therefore a productivity increase; 30% space reduction (parts/sqm); 100% quality check (i.e., all parts are 100% inspected); 0 NOK parts passed out of the production line; better quality results due to the dynamic positioning and better repeatability produced by using robotic systems; increased flexibility, because of the robotic system that can be programmed for more references; modularity. In conclusion, the fusion of ML, AI, robotic handling, and machine vision open a window for a new step in the industrial automation in the airbags/textile manufacturing, promising a productivity increase, a higher quality, less manual work, and better sustainability.