The headings correspond to the sections that we’re required to complete on the EPSRC form (http://www.epsrc.ac.uk/SiteCollectionDocuments/other/FinalReportFEC1207.doc ).
from the proposal.
The main objective of the SESAME project is to conduct high-quality scientific research to produce deployable systems that have a positive and measurable impact on the training of elite athletes. This project is application-led and multidisciplinary and has the following aims:
*) To understand coaching requirements by working with coaches throughout the lifetime of the project.
*) To support both athletes' and coaches' training and education through:
(i) capturing fine-grained data about athletes' actions and performance using wireless sensors, extensible middleware and networking platforms;
(ii) processing this data to extract information that is meaningful to coaches, using a newly developed biomechanical model to guide this;
(iii) identifying deviations between captured data and an idealised model of movement, supporting real-time corrective feedback;
(iv) storing the data and information in a long-term data store and performing trend analyses;
(v) visualising this information in a form that is meaningful to athlete and coach.
*) To explore the feasibility of correcting actions and building 'muscle memory' (the proprioceptive sense) through the use of real-time non-invasive wireless signalling.
*) To compare and evaluate different athletes' actions and performance, identifying advantages and disadvantages and informing customised feedback for individuals.
*) To place the UK firmly at the forefront of sporting technology.
The R&D process within the SESAME project is driven by a set of critical technical objectives. Consequently, the project activities have been structured as a set of interlinked milestones; these will be tackled by interdisciplinary interinstitutional teams, led by those with world-leading expertise. Details of this approach are given in the proposal proper.
We have developed a high-precision, high-frequency (1000Hz) wearable sensor technology to measure the foot contact times and step durations of athletes throughout a sprint or other intensive track or field event. The system and its associated analytic algorithms have been validated against ‘gold standard’ (force plate) data.
We have developed a high-frame-rate video system that can be operated by athletes and their coaches to monitor an athlete's motion alongside the above-mentioned sensor data.
Using these two systems we have collected a substantial body of data from elite sprinters during training runs and have used that data to derive new analysis algorithms. The datasets will be augmented and further exploited in future research.
We developed a system that can be worn by athletes and is designed to measure aspects of their movements. We investigated the use of cutting edge inertial sensors but ultimately discovered that pressure sensors in a shoe insole were able to measure many aspects of how a foot hits the ground and this in turn allowed us quantitatively assess athlete’s motions in new ways.
The foot-pressure sensor system was refined into a robust design and was deployed in collaboration with our UWIC SESAME partners on several elite sprinters at the National Indoor Athletics Centre (NIAC) in Cardiff to collect 30+ datasets. Each dataset records all of the footfalls of a sprinter in a 60 metre training run. The datasets were exploited to develop reliable algorithms for the extraction of foot contact times and step durations. In a validation exercise undertaken against force plate data with 15 of the datasets, the foot contact time obtained from the sensor system was found to correspond with the force plate to an accuracy of 1-2 milliseconds.
Our solution is both novel and highly adaptable. It does not require the trackside infrastructure that current systems demand; does not have long setup times; and is not noticeable to the athlete. It can be used for the measurement of foot contact times in almost any type of athletic activity. Its portability means it can be used in almost any training environment and over longer distances than previously possible. The data collection platform is entirely custom, developed specifically with sport in mind. Therefore it is very light, very small, and yet retains many data interfaces to allow expansion to use other sensors.
To complement this system we developed a data storage and visualisation architecture to allow coaches and athletes to review their data. This was augmented with a custom video system designed from scratch to support multiple, synchronised high speed video cameras and allow slow motion instant replay of sporting events-something that is not available even today, but is very popular with coaching. By integrating our wireless sensor system with this video system, we were able to provide intelligent in-field navigation through the video and to demonstrate novel visualisations.
In a parallel research strand, our early attempts to track athletes using on-body MEMS inertial sensors resulted in novel, prize winning research that allowed the tracking of pedestrians around buildings. The extra complexity inherent in sporting movements has so far prevented the application of the latter technology in sport, but the pedestrian research has laid the ground for this in the future.
- feasibility of a wearable sensor platform for use in monitoring motion of elite sprinters.
- feasibility of reliably extracting foot-contact intervals and step frequencies from low-cost and robust (FSR) sensors
- an algorithm for reliably extracting foot contact times (foot down and foot off) from FSR data, with comparable accuracy to the established sport science methods.
- validation of these technologies for a substantial set of data from multiple athletes.
- Custom wireless sensing platform that is small, light and extensible. It is better suited to sports than other BSN nodes
- Custom high speed video collection and review system that is in use by coaches and athletes today.
- Identification of new research avenues for technical sporting analysis and body sensor networks.
- Novel techniques to track humans using on-body inertial sensors
- Extensions to the Linux operating system to support a widely-used sensor platform have been made available as open source code and is in active use in the research community.
Journal paper:
* R. Harle, S. Taherian, M. Pias, G. Coulouris, A. Hopper, J. Cameron, J. Lasenby, G. Kuntze, I. Bezodis, G. Irwin, D. G. Kerwin " Towards Real-time Profiling of Sprints using Wearable Pressure Sensors ", Comput. Commun., vol. 35, no. 6, pp. 650--660, Elsevier Science Publishers B. V., Mar 2012.
Refereed conference papers:
* R. Harle, J. Cameron and J.Lasenby, “Foot Contact Detection for Sprint Training”, 2nd International Workshop on Video Event Categorization,Tagging and Retrieval (VECTaR2010), workshop of ACCV2010, November 2010.
de Boer, Willem and Lasenby, Joan and Cameron, Jonathan and Wareham, Rich and Ahmad, Shiraz and Roach, Charlotte and Hills, Ward and Iles, Richard , “SLP: A Zero-Contact Non-Invasive Method for Pulmonary Function Testing”, Proceedings of the British Machine Vision Conference, 2010, BMVA Press, pp. 85.1--85.12.
Oliver Woodman, Robert Harle, "RF-Based Initialisation for Inertial Pedestrian Tracking," Pervasive, series Lecture Notes in Computer Science, vol. 5538/2009, pp. 238-255, Springer Berlin / Heidelberg, May 2009
* Oliver Woodman, Robert Harle, "Pedestrian Localisation for Indoor Environments," Proceedings of the Tenth International Conference on Ubiquitous Computing (UbiComp 08), Seoul, Korea, ACM, Sep 2008
* S. Taherian, M. Pias, R. Harle, G. Coulouris, S. Hay, J. Cameron, J. Lasenby, G. Kuntze, I. Bezodis, G. Irwin, D. Kerwin , " Profiling Sprints using On-Body Sensors ," Proc of PerSens'10 - Sixth IEEE International Workshop on Sensor Networks and Systems for Pervasive Computing , IEEE, Mar 2010
Kuntze, G., M. R. Pias, Bezodis, I.N., Kerwin, D.G., Coulouris, G., Irwin, G. , " Use of on-body sensors to support elite sprint coaching ," International Association of Computer Science in Sport. Canberra, Australia , pp. 71-75, Sep 2009
* Kuntze, G., M. R. Pias, Bezodis, I.N., Kerwin,D.G., Coulouris, G., Irwin, G. , " Custom-built wireless pressure sensing insoles for determining contact-times in 60m maximal sprint running ," International Society of Biomechanics in Sport, Limerick, Ireland , Aug 2009
Rachael Casey , Adar Pelah , Jonathan Cameron and Joan Lasenby , Influence of step frequency on visual speed perception during locomotion, Poster paper, 7th symposium on Applied Perception in Graphics and Visualization , July 2010
Pias, M., Xu K., Hui, P., Crowcroft, J., Yang, GH., Li, V., Taherian, S. , " Sentient Bikes for Collecting Mobility Traces in Opportunistic Networks ," In Proc of the ACM SIGMOBILE 1st International Workshop on Hot Topics of Planet-scale Mobility Measurements (HotPlanet'09) , ACM Sigmobile, Jun 2009
Simon Fothergill , " Examining the effect of real-time visual feedback on the quality of rowing technique ," 8th Conference of the International Sports Engineering Association 2010 , Vienna, Austria, series The Engineering of Sport, Springer, Jul 2010
Simon Fothergill, Robert Harle, Sean Holden , " Modelling the model athlete: automatic coaching of rowing technique ," Joint IAPR Workshops on Structural & Syntactic and Statistical Pattern Recognition , Springer, Dec 2008
Simon Fothergill, Robert Harle, Brian Jones , "Context awareness for health and fitness : a case study", accepted for IWFAR 2011, June 2011
------------------------------------------------
Open hardware design: iMote2 WiFi board design :
This is the URL with the documentation, CAD files for the open-source release of the UCAM-WSB100:
http://sourceforge.net/apps/mediawiki/imote2-linux/index.php?title=UCAM-WSB100
Linux Operating System code :
The Industrial I/O subsystem of the open source Linux Kernel was developed developed within the SESAME project by Dr. J. Cameron as a solution to the problem of high sample rate sensor interfacing in embedded systems. The resulting framework has been available within the staging section of the main kernel tree since August 2008 and has since attracted contributions from Analog Devices, Nvidia, Nokia, TAOS and Texas Instruments alongside academic and non commercial contributions. The framework provides a unified interface with buffering and event handling capabilities. At the current time the 54000+ lines of code consist of the core largely written as part of the Sesame project and 54+ drivers supporting over 150 different device types. ( 77 ADCs, 12 Accelerometers, 6 ADC DAC, 24 DAC, 9 DDS, 8 Gyros, 9 IMUs, 5 Light sensors, 2 Magnetometers, 8 Energy Meters, 3 Resolvers) (6 drivers from SESAME).
The SESAME project also contributed updated support for the Intel Research IMote2 and Stargate2 platforms which have been further built on by other members of the community to ensure complete support for these common research platforms.
Discovery Channel TV presentation : http://watch.discoverychannel.ca/#clip404671
BA Festival of Science : public engagement
------------------------------------------------
Dr. Rob Harle, CUCL, Cambridge University School of Technology Strategic Grant for work in Sport Sensing, £46k, from March 2011. The University of Cambridge has recognised the possibilities for this work in the lead up to the 2012 Olympics and beyond. The School of Technology has provided a strategic grant to continue the work.
Dr. Joan Lasenby, CUED, EPSRC Knowledge Transfer (KTS) grant, ref. KTS EP/D076935/1 £42K, from 16 October 2010 for 1 year. For work in the application of some results from SESAME in for non-invasive lung function monitoring in collaboration with PneumaCare Ltd.
GloboSense - www.globosense.co.uk
PneumaCare - www.pneumacare.com
}}} === Contacts === At the Computer Lab: [http://www.cl.cam.ac.uk/research/dtg/www/people/rkh23/ Dr. Rob Harle], [http://www.cl.cam.ac.uk/research/dtg/www/people/gfc22/ Prof. George Coulouris] [[BR]] At the Engineering Department: [http://www-sigproc.eng.cam.ac.uk/~jl/ Dr. Joan Lasenby], [http://www-sigproc.eng.cam.ac.uk/~jic23/ Dr. Jonathan Cameron]