A method for collaborative knowledge acquisition and modeling enabling the development of a knowledge-based configurator of robot-based automation solutions

Abstract

When it comes to the development of knowledge-based configurators, there is a lack of procedures for the collaborative work on knowledge acquisition and modeling. However, the division of work is inevitable, especially in interdisciplinary application domains such as the configuration of automation solutions. Therefore, this paper presents a practicable method, enabling the collaborative development of a knowledge-based configurator across various partners and locations. The developed method is divided into eight steps: Firstly, economically relevant use cases are identified so that the necessary human and financial resources are released by the management. Secondly, several best practices, in this case successfully implemented robotic-based automation solutions, are collected. In the third step, generic, cross-case topological elements are extracted. Building on the relevant topological elements, constraints for the formation of the configuration logic are derived in the fourth step. In the fifth step, the required attributes are deduced from constraints and implemented within the data schema. The integration of the constraints, as well as the data schema in the configuration framework, takes place in the sixth step. In the seventh step, concrete components, e.g. individual end effectors, robot arms or controllers, are collected and prepared according to the data schema and integrated into the configurator. In the eighth and last step, the configurator is tested by comparing given configurations with known best practices. By deriving new constraints from the identified discrepancies and by analyzing new best practices, the resulting configurator can then be optimized iteratively. In addition, an auxiliary tool to support single steps of the method is introduced. For validation purposes, a suitable configuration model for robot-based automation cells was designed, distinguishing between scenario, product, processes and resources.