Assessing and predicting human performance using simulator data and probabilistic methods

Billard, Randy Joseph (2021) Assessing and predicting human performance using simulator data and probabilistic methods. Doctoral (PhD) thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (2MB)

Abstract

Lifeboat training is normally conducted in calm waters to minimize the risk to trainees and equipment. Practice in anything other than benign conditions is prohibited. Trainees are given little or no opportunity to practice in conditions that are probable in an emergency, including moderate sea states and reduced visibility. Coxswains are also expected to be able to deal with hazards and equipment faults although they are not exposed to these conditions in practice. Consequently, little is known about how trainees will perform in an actual emergency and the modeling of human performance in harsh environments has not been possible due to the scarcity of human performance data. With the advent of lifeboat simulator technology, it is now possible for trainees to practice in adverse weather conditions and to apply their skills in realistic emergency scenarios. Data can now be collected to assess how skills are acquired in training and how skills transfer to new operating scenarios. This data can be used to create models to investigate learning and to predict performance. The research in this proposal uses data collected from experimental studies performed with a simulator to study skill acquisition and retention, to predict human performance in emergencies, and to form models of competence that can be used to study this problem space. The thesis also provides insights on human performance and equipment limitation and uses numerical simulations to generate data of lifeboat launches into high sea states. The thesis comprises of four research papers, presented as chapters. The first paper evaluates how the type of training received affects the performance of lifeboat operators based on their ability to complete tasks in an emergency scenario. In the second paper, Bayesian inference is used to produce models of human performance to investigate skills acquisition in new trainees transfer of skills to new scenarios. The third paper presents a method to create models of competence using Bayesian Networks which are derived from expert prediction and experimental data. The final paper examines the performance of lifeboats in high sea states and the impact of coxswain timing on the launch performance, using data collected from numerical simulations. The contribution of the research is 1) knowledge on the amount of practice needed to achieve and retain competence to launch an lifeboat, 2) an evaluation of how skills acquired in training transfer to new scenarios, 3) knowledge on how the type of training received affects performance in an emergency scenario, 4) insights on how much practice is needed to learn different lifeboat task types, 5) an increased knowledge of equipment performance limitations in weather conditions possible in an offshore emergency, and 6) methodologies to create probabilistic models of performance that can be used to study learning and adapt training. The study outcomes have relevance to training providers and presents methodologies that can be used to study other problem areas. The scope of work is performed in five studies using the outcomes of a human factors experiment and numerical simulations.

Actions (login required)

View Item