Postgraduate research project

Combining embedded sensor technology and machine learning to quantify swimmers’ performance in a normal training environment

Funding
Fully funded (UK only)
Type of degree
Doctor of Philosophy
Entry requirements
2:1 honours degree View full entry requirements
Faculty graduate school
Faculty of Engineering and Physical Sciences
Closing date

About the project

Coaching feedback systems in competitive swimming are limited by the quality of the data available, the time taken to set up equipment and post process data. This severely limits the number of training sessions that can be analysed in detail making it hard to accurately assess a swimmer’s training load and determine how their technique could be improved in the future.

This project, sponsored by British Swimming and the English Institute of Sport, will use small unobtrusive sensors embedded in swimming apparel to measure key performance metrics during normal coaching sessions. This will allow quantitative data to be obtained and processed quickly and automatically to assist coaches and biomechanists to improve athlete performance. 

Embedding the sensors into normal swimming apparel minimises the impact they have on the athlete’s technique, however current sensors of this size with adequate power supply requirements and efficient data transfer methods are not available and will require some novel approaches and development work. 

Machine learning techniques will be used to automatically classify the data into predefined activities based on the typical phases of a race and the different competition strokes. The classified datasets will allow macro-level performance indicators such as time spent in different phases and mean swimming speed per length, allowing a quantifiable measure of training load to be determined.

The ultimate objective is to then develop automated performance analysis at a stroke-by-stroke level to determine key performance indicators. Initially data will be collected using club-level swimmers with currently available sensors to allow a large, unique, data set to be obtained with varying levels of athlete ability before embedding these systems into British Swimming training sessions.

These datasets will allow the complex interactions between individual swimming kinematics and overall athlete performance to be analysed using AI-based techniques.

This research project is supported by British Swimming and the English Institute of Sport who have collectively supported over 15 PhD projects working with Olympic athletes within the Performance Sport Engineering Lab at the University of Southampton.