An Investigation into Kicking in Women’s Australian Football

[thumbnail of CUST Emily-thesis-redacted.pdf]
Preview
CUST Emily-thesis-redacted.pdf - Submitted Version (16MB) | Preview

Cust, Emily ORCID: 0000-0001-6927-6329 (2020) An Investigation into Kicking in Women’s Australian Football. PhD thesis, Victoria University.

Abstract

In Australian Rules football (AF), kick skill performance involvements, notably the drop punt, are statistically strong contributors towards team match success. The start of a National women’s AF competition (AFLW) in 2017 created opportunity for new knowledge to be established around the characteristics of AFLW athletes’ skilled performances. Using developments in inertial measurement unit (IMU) technology and analytical methods, this thesis takes a multi-disciplinary approach to analysing AFLW skilled performances and subsequently proposes a concept of a semi-automated AF kick type classification system for skill monitoring in an applied environment. Specifically, the thesis: 1) evaluates the research literature on machine learning for sport-specific movement recognition, 2) determines the importance of AFLW athlete skilled performance indicator contributions during match play, 3) defines AFLW drop punt kick kinematics, and 4) evaluates AF kick type classification models using IMUs as a proof-of-concept to support further developments in the area. Understanding analytical methods previously implemented with IMU or computer vision data and the evaluated capacity of these models in sport-specific movement recognition literature, is important in the adaptation for, and application towards new problems in sport. The first part of this thesis focuses on the experimental set-up, data pre-processing, and model development methods in the relevant literature on recognition of sport-specific movements in-field using IMU or computer vision technology. Of the 52 studies identified, 29 used IMUs, 22 used vision data and one study integrated both technologies. Supervised machine learning models were the dominant approach for developing sport specific movements recognition systems. Although nine studies implemented deep learning algorithms which comparatively indicated superior results to machine learning models, and demonstrated the advantages and potential of these model types. This study also highlights the importance of considering the model and overall system development in relation to the targeted sports movement(s) when progressing future research in the field. The applications of IMUs for sport skill recognition and subsequently performance analysis in-situation demonstrated in the literature may be beneficial in AF. As AF matches are technically skilled in nature, this thesis sought to investigate relationships of AFLW athlete skill performances in explaining team quarter and match success which knowledge was previously limited. Performance indicator distributions in explaining match quarter outcomes show the strongest skilled contributions from key high performing athletes, and the overall team strongest features related to kick performance indicators. Considering the importance of the kick in AF, the thesis then continued to define the kinematics of AFLW athlete’s drop punt kicks across leg preferences which was unknown. Several key differences from men’s AF kicks were found, also, women’s kick movement patterns quantified which is beneficial for specific coaching practices. Developments in IMU use for sport-specific movement recognition through machine learning models demonstrate advantages in sporting performance analysis applications. In the final section, these technological developments are investigated for the concept of a semi-automated AF kick monitoring system using IMUs. The work is applied in an AFLW training environment as a unique study for capturing the importance kick skill performance towards team match success and differentiation from men’s AF kick biomechanics. The findings indicate that kick types can be sufficiently distinguished from one another which creates scope for further applied work in AF training sessions. Overall, the work in this thesis is the first to establish the biomechanical characteristics of elite women’s AF kicks and enhances the knowledge of skilled performances in the AFLW. Furthermore, it is the first to implement IMUs for on-field AF kick recognition. Increasing automation in sport-specific movement recognition can be applied in AF kick skill monitoring; particularly as a unique forefront in AFLW sport science applications towards kick performance improvement. The methods used and findings of this thesis can also be transferred to other elite women’s team sporting leagues involving kicking actions such as Rugby and Gaelic football.

Additional Information

This thesis includes 1 published article in the appendix for which access is restricted due to copyright. Details of access to this paper has been inserted in the thesis, replacing the article themselves.

Item type Thesis (PhD thesis)
URI https://vuir.vu.edu.au/id/eprint/41271
Subjects Historical > FOR Classification > 1106 Human Movement and Sports Science
Current > Division/Research > Institute for Health and Sport
Keywords Australian Rules football; inertial measurement unit; AFLW; drop punt kick; performance; biomechanics; machine learning; deep learning
Download/View statistics View download statistics for this item

Search Google Scholar

Repository staff login