IAA Implementation of a marker-less motion capture system for skeleton push-start performance analysis

Across the last three Olympic cycles, the British Skeleton team have won 3 Olympic Gold and 2 Olympic Bronze medals becoming the most decorated nation in this sport. Across these preparatory cycles, the team have worked with the University of Bath on various research projects focussed on enhancing the preparation of their athletes for the very top level of competition. Dr Steffi Colyer has been involved in all projects to date and is working closely with the organisation to develop future research priorities. This project portfolio could be formed into a future REF Impact case study for the next REF.

The latest project has been conducted within the EPSRC funded Centre of the Analysis of Motion, Entertainment Research and Applications (CAMERA) and has been a close collaboration between the investigators listed above and the BBSA (a founding partner of CAMERA). As part of this project, we have developed novel motion analysis technologies that allow biomechanical metrics and British athletes’ performances to be assessed in a totally non-intrusive way at the dry-land push-track situated on the University of Bath’s campus. This has been a central theme of work within CAMERA and we have several papers relating to developing this (see section 7 below). This fully passive assessment capability has never previously been achieved and is at the forefront of this field.

Specifically, traditional motion analysis requires the placement of reflective markers on participant athletes to track their motion. This is a time-consuming process and creates an unrealistic performance scenario. Most critically, this assessment procedure is not feasible in the normal training or competition setting of skeleton athletes (i.e. ice-tracks), limiting the performance information available to coaches and performance staff. University of Bath researchers from the Departments of Health and Computer Science have recently developed and validated a marker-less motion capture system, which does not require the attachment of markers or any equipment, and instead extracts and synthesises information from regular video using advanced computer vision algorithms. However, the impact of this system cannot yet be fully realised for athlete training, as advanced expertise and lengthy data processing are

currently required to provide usable outputs.

This project will assign a dedicated RA to engineer the system for deployment, making it fully implementable to assess everyday training and preparation of athletes. This will include the verification of the system’s utility in a realistic (ice-track) setting, at a European ice-house (Cesana). Success will allow us to impact the BBSA’s routine high-performance training and assessment practice in a way that was previously unobtainable. By the end of the proposed project, we will have created a system that can be used independently by non-experts (i.e. BBSA coaches and performance staff) to obtain unique and valuable data to inform the training and preparation of athletes in the lead up to the 2022 Winter Olympics in Beijing and beyond. Ultimately, this system could be deployable across other sports and generate significant research impact while providing competitive advantage for the UK in several elite sports.