Phd Proposal : Sensors for Quantitative Sports Performance Analysis
Introduction
Performance analysis systems for sports such as football use multiple cameras to track players throughout a game. From the data gathered, tactics can be evaluated and statistics such as the total distance run by a player can be used to analyse each player’s performance.
For performance analysis in sports such as athletics we are interested in tracking human movement to a much higher degree of precision. Knowing the angles of an athlete’s joints and the positions of various parts of their body throughout a performance can be used to analyse their technique and provide them with quantitative feedback.
Existing Solutions
At present there is no single satisfactory solution to the problem of how to accurately measure human movement during functional activities. Devices such as electrogoniometers can be used to accurately measure joint movements; however they are not suited to sports analysis since their attachment will affect an athlete’s performance.
The most common approach for tracking movement is to perform kinematic analysis using filmed footage, from which quantitative data such as joint angles can be obtained using computer vision techniques. There are, however, a number of problems with this approach. The occlusion of markers attached to the athlete or simply the occlusion of body parts is a major problem. The computer vision algorithms necessary to recover quantitative information are also computationally expensive and often need manual assistance in order to achieve accurate results. These issues present real problems at a competitive event. There may only be a limited number of positions from which an event can be filmed and an athlete may only have a short amount of time between successive attempts in order to seek feedback from their coach. Ideally we would like accurate, reliable and complete data available in real time
- or at least immediately after - an athlete’s performance.
Sensors for Performance Analysis
I propose to investigate the use of sensors attached directly onto an athlete to provide reliable and high precision tracking of their movements during an event. UWB tag technology can be used to track small tags at a high time resolution to an accuracy of 15cm. Post-processing algorithms can be used to reduce the positioning error to around 5cm. Further advances in this field could facilitate the use of UWB tags for tracking human motion to a higher degree of precision. An alternative approach could be to use proximity sensors. Distances between multiple sensors attached to an athlete could be used together with the constraints the human body to infer the relative positions and angles of body parts and joints.
A number of problems must be solved before sensors can be used to accurately measure human movement during functional activities. I propose to explore potential solutions to these problems.
• How can sensors be attached to an athlete in a way which does not affect their performance, and from which reliable data can be obtained?
• What sensors and algorithms can we use to track human movement to a high level of precision?
• If proximity sensors are used, how can the positioning information be transferred from the athlete to a laptop for analysis? Ideally we would like to transmit the data in real time, particularly for events which last for some time such as long distance running. What protocols can we use to do this given the amount of data we need to transfer and the size and power constraints imposed on the sensors?
• Once data has been gathered what type of analyses can we perform in order to provide the coach and athlete with useful information? How can this information be best presented?
To investigate possible solutions for these problems I will first have to conduct detailed research into existing solutions, their problems and the technology used. It is anticipated that this will take approximately half of my first year. This research phase will allow me to identify the most promising existing techniques, which can then be used as a starting point for further research and development. It will be necessary to develop and test possible
solutions, using trials with athletes to provide feedback.
Benefits
A system which can track human motion to a high degree of accuracy will have several uses. In athletics the most obvious benefit is to the athlete and coach, who can use the data to identify tiny changes in technique which can have a major effect on performance. It may be possible to identify patterns which lead up to a peak in an athlete’s performance. An attempt to replicate such patterns in the build up to major events could then be made. The data could even be used to attempt to spot when an athlete is likely to incur an injury. Being able to accurately track human movement also has many other potential uses not related to sport. For example such a system could be used to analyse the movement of individuals using prosthetic limbs in order to improve their design.
