A method for modelling operator behaviour applied to the NewTranspall microworld

A method for modelling operator behaviour applied to the NewTranspall microworld

Nap, R.

Wieringa, P.A. (mentor)
De Winter, J.C.F. (mentor)

Mechanical, Maritime and Materials Engineering

Biomechanical Engineering

2007-03-08

Violation of safety procedures by operators is an important cause of accidents. To prevent violations effectively a thorough understanding of the operators' decision making process is required. This research proposes a method to create a personal model that predicts the action the operator would take based on the state of the system. We strive to create a model that reproduces as many operator actions as possible. Decision data used for this research has been collected during experiments in which nine subjects operated the NewTranspall microworld. In developing the model, first the researcher defines the likely decision criteria (i.e. in this research safety, productivity and workload based on prior research). Next, expert subjects are asked to rank screenshots of situations according to the criteria during an interview. Their explanations for this ranking reveal the key parameters that define the decision criteria on which they base decisions. The developed model valuates all possible actions at the system state when the operator took an action; the criteria values are calculated using the key parameters. The model then combines the criteria values using multi-criteria utility theory to form a preference value for each possible action. The highest preferred action is the predicted action. The models are then fitted to the decision data to maximise the amount of correct predictions. It is concluded that the method is successful in a relatively simple NewTranspall environment. A subject's cognitive map visualises the subject's preference, allowing for easier insight into their decision making.

human decision making
microworld
modelling
violating behaviour

http://resolver.tudelft.nl/uuid:27413819-4ad0-4bcf-b884-d3009a4c7dbb

Student theses

master thesis

(c) 2007 Nap, R.