A Novel Skill Modeling Approach: Integrating Vergnaud's Scheme with Cognitive Architectures

Abstract

Human-machine interaction is increasingly important in industry, and this trend will only intensify with the rise of Industry 5.0. Human operators have skills that need to be adapted when using machines to achieve the best results. It is crucial to highlight the operator's skills and understand how they use and adapt them [18]. A rigorous description of these skills is necessary to compare performance with and without robot assistance. Predicate logic, used by Vergnaud within Piaget's scheme concept, offers a promising approach. However, this theory doesn't account for cognitive system constraints, such as the timing of actions, the limitation of cognitive resources, the parallelization of tasks, or the activation of automatic gestures contrary to optimal knowledge. Integrating these constraints is essential for representing agent skills understanding skill transfer between biological and mechanical structures. Cognitive architectures models [2] address these needs by describing cognitive structure and can be combined with the scheme for mutual benefit. Welding provides a relevant case study, as it highlights the challenges faced by operators, even highly skilled ones. Welding's complexity stems from the need for constant skill adaptation to variable parameters like part position and process. This adaptation is crucial, as weld quality, a key factor, is only assessed afterward via destructive testing. Thus, the welder is confronted with a complex perception-decision-action cycle, where the evaluation of the impact of his actions is delayed and where errors are definitive. This dynamic underscores the importance of understanding and modeling the skills of operators.