Al toolbox for Lot-Site-One robot welding

The AIT00LS1 project deals with new applications of novel AI technologies for robot welding, such as AI tools for process control of complex multi-parameter dependent systems, as well as manufacturing quality control. Specific implementations include improving the weld quality by introducing ML models capable of dynamically optimising the robot welding processes.
However, in industrial applications the problem withdata-based AI is the acquisition of large and reliable dataset(s). In several manufacturing applications, the size of the dataset is limited, which is a major limitation for the use of traditional AI methods. The data scarcity prolan is particularly evident in small series and Lot-Size-One (LSI) production. Hybrid modelling involving measurement data but also other form of knowledge capable of capturing physical phenomena and know-how is a solution to these challenges. These developments apply to other manufacturing methods, eg. Additive Manufacturing AM. too.

For the participating companies, the key project topic is the effect of robot welding procedure on resulting weld quality and process productivity. The AI model teaching data include, eg. welding parameters, measured geometrical weld characteristics and global dimensional accuracy. The final company project goal is an AI based autonomous online process and quality control system for automated Lot-Size-One welding production. AI based production and quality control processes in manufacturing also enable full product traceability through the whole production chain and provides possibilities for extension over the whole lifecycle for eg. optimising the service and maintenance operations. For the Original Equipment Manufacturers (OEM) the main project impact is the improved market position due to productivity increases and enhanced quality and consistency, leading into reliable long-term service properties.