HCC: Medium: A Multi-disciplinary Computational Model of Worker, Task, and Context for Real-time Adaptive Procedural Systems (R-TAPS)

This project will advance fundamental knowledge of how to consider psychological, contextual, and technical issues in designing tools to help workers in safety-critical systems follow work procedures and reduce human errors. One of the most frequent causes of incidents in high-risk facilities is when workers deviate from procedures. Even well-designed procedures and tools for following them are not immune, because of a number of factors including workers’ experience with the task, how often they perform it, and how well the procedure design fits their cognitive style.

Together, these findings indicate that rather than single, fixed procedure formats, better safety is likely to come when procedures adapt to the person and the context they are used by. To do this, the project team will bring together expertise from psychology, human factors, and engineering to develop models of the important factors that affect people’s performance around procedures, and to design procedures based on sets of behavioral “scripts” that can be activated in appropriate situations and tuned to particular workers.

The work will be done in the context of the oil and gas industry, a domain that has both high societal value and high safety risks, with the goal of improving worker safety in real deployments in this domain along with general methods that can be applied to studying and designing safety procedures in other safety-critical domains. The team will also provide career development opportunities for several students, with a focus on increasing participation of underrepresented groups in research.

To do this work, the project team will leverage cognitive, linguistic, and systems methods to identify worker, task, and context attributes relevant to procedure performance. This project's specific research objective is to map these factors onto a computationally tractable framework, the Multi-disciplinary Interactive Behavior Triad, that will form the basis of a script-based Real-Time Adaptive Procedure System (R-TAPS) that uses the model and data collected from work context, matching a worker’s current context to known factors and design features and automatically generating personalized procedural guides.

Data for both building and evaluating the models and the R-TAPS system will be collected from workers recruited at Shell's Robert Training Facility (RTF) with a high-fidelity simulation of offshore operations to test and evaluate the R-TAPS tool. More generally, the research contributes to human-machine interaction by creating a framework that allows for methods from multiple disciplines associated with human behavior to be integrated and applied to an AI methodology, where all the relevant constructs can be applied and translated to the design of effective technology.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.