Resource type
Thesis type
(Thesis) Ph.D.
Date created
2023-12-11
Authors/Contributors
Author (aut): Mayerhofer, Patrick
Abstract
Many athletes do not have access to a coach to help them improve their performance, and coaches' feedback can be subjective. Artificial sport coaches can collect continuous data during training and competition, analyze these data, and — by comparing the current performance with a target performance — give automatic feedback. Artificial sport coaches can make sports more accessible for athletes without coaches, and help coaches make more objective decisions. Developing artificial sport coaches can be challenging, as they require knowledge of continuous data collection outside of the laboratory, smart data analysis, and control systems. The goal of this thesis was to develop and test tools that can simplify the process of gathering objective data during sports, analyzing these data, and providing automatic real-time feedback to the athlete and the coach. First, my colleagues and I designed a generalizable approach to build a closed-loop feedback control system in sports, and tested this approach in controlling cycling power. Second, we demonstrated how a data-driven approach can simplify the process of developing complex models, by comparing the advantages and disadvantages of physics-based and neural network-based modeling for predicting cycling power. Third, we tested whether we could use a state-of-the-art image recognition to classify individual runners and their running performance. And lastly, we demonstrated how we can teach important, complex laboratory skills needed for human data collection and analysis, without the need of a physical laboratory or its expensive laboratory equipment, by utilizing the strengths of wearable sensors for remote teaching. Overall, my thesis provides emerging and established scientists and sports and health technology engineers with a better understanding of how to more easily and efficiently develop wearable measurement technologies, data analysis systems and complex models, and control systems. Additionally, it will also provide novel sports specific insights, specifically for cycling and running.
Document
Extent
102 pages.
Identifier
etd22879
Copyright statement
Copyright is held by the author(s).
Supervisor or Senior Supervisor
Thesis advisor (ths): Maxwell, Donelan, J.
Language
English
Member of collection
Download file | Size |
---|---|
etd22879.pdf | 4.77 MB |