LAUSR

Abstract Due to globalization, producing companies are facing competitive pressure. The results are a growing number of product variants, lower prices, higher product quality and shorter life cycle times. Due to the rising complexity in process planning and product assembly operation - e.g. the product assembly - cognitive assistance systems are increasingly used especially in manual assembly. These systems ease the work of the employee on the shop floor and increase quality. Additionally, the employee can handle a wide range of variants. These assistance systems are often combined with inspection systems such as 2D-cameras, the information is provided and a process verification is performed. These systems are particularly helpful in the rework area, where the assembly processes depend on the variant and on the detected error to be fixed. Still, current systems require the employee to follow exactly the instructions without any deviations from the specified path. This deviation can occur, especially during repair if the employee goes backwards in the process because he assumes another cause of error than presumed from the test bench. This article presents an assistance system, which reacts flexibly to a specific error and acts situational to deviations. During manual assembly or repair, several process indicators are checked simultaneously, so that the actual process state can be detected. A certainty must be given to react situationally on deviations from a given plan. It is also considered that not all sensors can reliably detect the current product or process state. The paper is structured as follows: introduction, initial situation, approach, validation and conclusion.