Self-optimising Assembly Systems for Handling Large Components

In the field of assembly planning, optimisation approaches are often limited to partial evaluations of the value creation chain due to complex interactions between the components of the production system. The usage of situational adaptive systems helps to reduce the risk of overly focusing on individual elements without considering side-effects. Especially integrative, self-optimising structures offer great potential for improved planning efficiency. In this research a three-layered assembly planning model was established and implemented. The developed software structure includes a hybrid approach with offline planner, conducting all preliminary analysis with an assembly-by-disassembly strategy, and online planner, evaluating this information during the assembly to derive a suitable sequence for the current production situation. Furthermore a cognitive control unit is responsible for the decision-making and executes appropriate actions. For validation, a robot-supported assembly cell is presented. Two series of experiments were conducted to develop a concept that adapts the system behaviour to operators' expectations by using human-centred process logic. Additionally a lab study was designed to investigate the visual presentation of information to humans. The work achieved a scientific examination of cognitive mechanisms in automation. It shows that cognitive automation of production systems enables an efficient and robust assembly of diversified product families. This effectively makes customer-oriented mass production possible and offers high-wage countries considerable competitive advantages.