AI-Human Collaboration for improved Situation Awareness in the maintenance domain

The goal of this project is to explore the human abilities to develop Situation Awareness of the changing situations of engineering systems in order to facilitate maintenance and make recommendations to improve Situation Awareness about intelligent maintenance systems.

Sponsors: Trafikverket/JVTC

Researchers: Prasanna Illankoon (PhD candidate), Uday Kumar, Phillip Tretten (PL)

Duration: 2016-2020

Background

The use of intelligent systems has resulted in higher system reliability, a higher quality product, and reduced risk for human error. One key aspect concerns the fact that human interaction has decreased although has become more critical than before. Human intervention is especially crucial when dealing with complex and safety critical systems, where and when immediate interventions are required. However, with the rise of intelligent systems, humans have lost their situation awareness and a good situation awareness is needed in dynamic environments if they are to intervene quickly and successfully.

Goal

The goal of this project is to explore the human abilities to develop Situation Awareness of the changing situations of engineering systems in order to facilitate maintenance and make recommendations to improve Situation Awareness about intelligent maintenance systems. Project answers three questions: What are the major information requirements related to the task, human, and environment that influence a human’s Situation Awareness during maintenance actions; What kinds of strategies do humans use to be aware of the necessary information for actions; What are the most effective strategies to provide better Situation Awareness to facilitate maintenance of future engineering systems. The main methodology used in the project is Cognitive Task Analysis. It involves Semi structured interviews, participatory observations, focus groups and case studies.

Project Status and Results

The project developed and tested five systematic methodologies to find suitable interventions to fulfil the situation awareness requirements. Case studies focused on situation awareness requirements during maintenance execution in manufacturing, aviation and railways. Project resulted in the identification of seven key situation awareness requirements for maintenance: detection of abnormalities; diagnosing and predicting their behaviour; making changes in system configuration; compliance with maintenance standards; conducting effective maintenance judgements; maintenance teams; and for safe maintenance work. Five strategies to maintain situation awareness were identified: explicit knowledge status, sense making, recognition primed decision making, skilled intuition, and heuristics. Project also showed why intelligent maintenance systems will make it difficult for humans to use most of these strategies to maintain situation awareness in future. Project showed that in the maintenance domain, keeping humans in the loop requires a novel collaborative approach where the collaboration of the strengths of intelligent systems and human cognition is necessary. Finally, a new theoretical model for decision support (Distributed Collaborative Awareness Model) was developed. The study also shows how to apply the model using Augmented Reality technologies in the industries including the railway maintenance sector. The candidate successfully defended the PhD at the end of the project.