The framework ADAPT, performs the automatic diagnosis of industrial automated processes (machines and operators). By interpreting, from process and operator data, the context and the fulfilment of procedures associated with tasks or operations: (i) verification of correct task performing with respect to a predefined ideal procedure and worker experience; (ii) novice operator training and guidance, increasing awareness, comfort, trust, skills and safety (physical and social) of workers.
The basis is a domain-independent metamodel inspired by the cognitive process that real domain experts or tutors (e.g. in educational scenarios) carry out while supervising or tutoring a learner during a process: experts first perceive facts or observations happening in the environment. Then, they interpret those observations in order to identify actions (i.e. they unconsciously assign meaning to their observations in the context of the domain) being carried out, and once they have processed this information, they make a diagnosis, which involves detecting and/or correcting mistakes.
The metamodel is divided into three subsystemas (a) Observation, (b) Interpretation and (c) Diagnosis subsystems.
Observation level: This level contains the necessary elements for specifying the observable facts that are interesting from an educational or process diagnosis point of view.
Interpretation level: Interpretation level describes generically how to recognize ISs with such accuracy that the diagnosis subsystem is able to determine whether the actions are correct or incorrect and its reasons. To that aim, it is essential to know the context where the actions are happening.
Diagnosis level: The objective of this level is to manage the elements that will allow an appropriate diagnosis of the activity to be generated. ADAPT provides the educational or process evaluation components of the built system with real-time diagnostic results or final evaluation results as needed. The diagnostic results contain information at different levels: the correctness of situations, the correctness of every step within each situation and other information such satisfied and non-satisfied conditions within the steps.
For more information, please contact:
- Luka Eciolaza – leciolaza@mondragon.edu