This work presents machine learning based techniques for detecting mind-wandering and predicting hazard response time in driving using only easily measurable driving performance data (speed, horizontal and frontal acceleration, lane gap, and brake pressure). Such predictors are relevant as research tools in the driving simulation community. We present a simple method, and a feature extraction based method, of representing time-series driving performance data that both support machine learning based predictions. We use the two types of representations to compare the effectiveness of support vector machines, random forest, and multi-layer perceptrons on data from 117 drives performed by 39 participants during a previous study in the high-fidelity driving simulator at the University of Guelph.Classification of mind-wandering and prediction of hazard response time was successful when compared to baseline measures. Specifically, random forest methods were most effective in both types of prediction and feature extraction supported the strongest random forest prediction of hazard response time. A discussion of the reasoning for this is included.To our knowledge this is the first driving pattern based classification of mind-wandering in a fully immersive driving simulator.iii
AcknowledgementsBefore anything else I wish to express my deepest gratitude to my amazing supervisor professor Andrew Hamilton-Wright. Andrew, your contributions to my academic journey cannot be overstated, the courses you taught me at Mount Allison are what inspired me to pursue this subject at the graduate level and the phenomenal quality of your support and mentorship here in Guelph have prepared and motivated me for my journey into the future.You are truly one of the best professors, leaders, and guides I have had the good fortune to meet.I would also like to express my deepest thanks to my co-supervisor professor Lana Trick.Lana, I want to thank you for introducing me into the world of driving simulation through the well organised, engaging, and collaborative meetings and events of the DRiVE lab, for your incredibly helpful feedback on my research project from design to completion, and for all the work you have put in over the years building and supporting the impressive lab that made this project possible.I gratefully acknowledge the assistance of my advisory committee member professor David Calvert for his time, insightful questions, and feedback. I also wish to thank Heather Walker, an incredibly conscientious PhD student in the DRiVE lab who trained me on the simulator when I arrived, provided consistent support to all of us in the lab, and, along with professor Trick, is responsible for the experimental work that this thesis relies on.Finally, I would like to thank my family, for their continued support through this degree and proofreading many drafts of many papers, especially my mother who has moved mountains for us for as long as I can remember.