enabling technologies for robust performance monitoring

Pre-published text for ISEA conference

ENABLING TECHNOLOGIES FOR ROBUST PERFORMANCE MONITORING Laura Justham1*, Sian Slawson1, Andrew West 2, Paul Conway 2, Michael Caine1, Robert Harrison 2,

(1) : Loughborough University Sports Technology Institute Loughborough Science and Enterprise Park 1 Oakwood Drive, Loughborough LE11 3QF, United Kingdom Phone: +44 (0)1509 564812 E-mail : {L.Justham, S.E.Slawson, M.P.Caine}@Lboro.ac.uk

(2) : Wolfson School of Mechanical and Manufacturing Engineering, Loughborough University, Loughborough, LE11 3TU United Kingdom E-mail : {A.A.West, P.P.Conway, R.Harrison}@Lboro.ac.uk

TOPICS: Sailing/Water Sports, Modelling Abstract: Monitoring and performance analysis of elite swimmers during training and competition is an under-developed area of research due to difficulties in collecting data while the athletes are in the water. A monitoring system which can provide timely feedback to the swimmer, coach and sports scientist regarding the performance and physiological capabilities of an athlete is critical for the development of optimised personal training plans that could ensure a swimmer’s continued improvement and enhance their ability to win medals in major competitions. This paper is focused on the development of the requirements of such a monitoring system. A critical review of current monitoring systems that are in use in swimming training has been carried out and has identified areas where technology is currently being used. These technologies include video analysis, pressure and force measurements, velocity and acceleration measurements and physiological monitoring. Suitable technologies for a novel integrated and distributed monitoring system have been identified and a prototype demonstrator, which is as non-invasive and nonencumbering as possible, is currently under development at Loughborough University. This prototype provides both inter and intra stroke information as well as video augmentation and a comprehensive web-based training database to provide a comprehensive monitoring system for swimming. Keywords: Swimming, Performance Monitoring, Sensor Networks, Image Processing

1- Introduction Sports monitoring systems which provide timely feedback to a coach and their athlete are critical at an elite level. Athletes require information regarding their technique, physiological capabilities and past performance in competition to develop a personalised training plan that is focused upon continued improvement in their chosen discipline. To this end the expertise from different professional practitioners, such as sports scientists and physiologists, is required to address the needs of each individual athlete and quantitatively assess their progression. A large number of commercial monitoring systems are available to support these activities in various sporting disciplines. The systems which are available focus upon both direct (i.e. equipment attached to the athlete) and remote (i.e. equipment remote to the athlete) monitoring techniques. The most commonly used direct monitoring techniques are sensors attached to the athlete. Inertial measurement units (IMU) which contain combinations of accelerometers, gyroscopes and magnetometers can be used to track human limb movement (Luinge 2002; Luinge 2005; Rootenberg 2006; Xsens 2008) and in outdoor sports GPS receivers can be used to measure details such as position, velocity and distance travelled (Garmin 2008). Data collected from these devices may be logged using on-board memory or transmitted wirelessly to a host computer and analysed with proprietary software.

-1-

PRE-PUBLISHED TEXT FOR ISEA CONFERENCE


The most commonly used remote monitoring techniques are digitisation of video, motion tracking using vision systems, force plates and pressure mats. Video digitisation requires manual post-processing of video sequences (Dartfish 2008; Quintic 2008).

-2-



Vision based motion tracking requires the athlete to wear reflective markers to allow their movement to be accurately tracked (Vicon 2008). Force plates and pressure mats are often wired directly to a computer to provide real-time feedback to the athlete regarding the force exerted during a predefined movement (Kistler 2008; RSScan 2008). These systems can provide informative and quantitative information however they can be limited by their set-up time and they often require the athlete to perform in a constrained environment such as in the laboratory. Both direct and remote monitoring tools are widely used in many sports however it is unusual that they are used in swimming due to the harsh environment where the athlete competes. It is the purpose of the current research to identify and develop an integrated monitoring system which is suitable for use in water. In order to carry this out, a critical evaluation of the current state of the art monitoring systems in swimming has been carried out and is presented in the following section. The prototype demonstrator, which is under development at Loughborough University, is being designed to deliver non-invasive, non encumbering real time feedback to the athletes and coaching personnel during both training and competition scenarios. Preliminary specification and design of the integrated monitoring system have been detailed in this paper.

2- Current State of the Art Performance Analysis Technologies A detailed literature review has been conducted to identify the current state of the art systems used for performance analysis in swimming. Seven types of monitoring have been identified, see Figure 1, and are (i) video analysis, (ii) physiological analysis, (iii) pressure or force analysis, (iv) velocity or acceleration analysis, (v) measurement of drag, (vi) ergometer analysis and (vii) theoretical analysis. These techniques have been divided into three distinct measurement approaches: 1. Remote Analysis where the required measurement equipment is separate to the athlete in the water 2. Direct Analysis where the required measurement equipment is attached to the athlete in the water 3. Modelled Analysis where testing is undertaken out of the water or where theoretical modelling is used to analyse the athlete Video or image processing is a popular remote performance monitoring technique as it can be used in both training and competition. Visual information is gathered and analysed using post-processing and digitisation. Qualitative analysis can be undertaken to discuss the athlete’s perception of their performance, whereas quantitative analysis can be performed by manually or automatically digitising images to allow information such as velocity, joint angle and body position with respect to elapsed time to be derived. To ensure that meaningful quantitative results are obtained, video hardware must be calibrated carefully to enable them to interface with software applications. Manual digitisation is currently undertaken using software such as Quintic (Quintic 2008) and Dartfish (Dartfish 2008) and automated analysis is possible for elements of the stroke, for example when considering the glide (Sanders 2007). In addition to digitisation and video analysis software, human motion capture can also be carried out using systems such as Vicon (Vicon 2008). Markers are applied to the swimmer’s body and tracked through the water that might inhibit or interfere with their swimming stroke. Direct monitoring forms the bulk of current analysis techniques in swimming and involves the swimmer wearing markers or equipment or being tethered to monitoring equipment on poolside. Velocity has been directly measured using tethered systems such as the commercially available SPEED system which was developed by AP Labs (APLabs 2008) and in research (Dekerle 2002). The swimmer is attached to a light cable and asked to swim as normal. The cable is tethered to a poolside tachometer, which provides a measurement of velocity. It should also be possible to integrate accelerometer data to calculate velocity; however this is felt to be an unreliable method under normal circumstances due to errors accruing from the integration process (Mackintosh 2008). Generally, information gathered from accelerometers is used to develop an understanding of stroke technique by positioning the devices on the wrists (Ohgi 1999; Ohgi 2002) or on the small of the back (James 2004). Specific and detailed information over multiple stroke cycles may be logged and represented graphically to visualise the movement of an athlete through space. The method of using accelerometer devices in swimming is still in its infancy in terms of development and complexity, however there is the potential for these types of devices to be developed further such that more thorough and detailed analysis techniques to monitor and understand stroke technique can be developed. Pressure and force sensors are used widely in land-based applications to measure the forces exerted by the body during various forms of exercise. Within water, the measurement of pressure or force becomes more complex due to the effect of the water. AP Labs have developed a hand pressure measurement system, KZ (APLabs 2008) which uses a differential pressure transducer to monitor the pressure applied by each hand as the swimmer moves through the water. This device requires the swimmer to wear a waterproofed pressure transducer on the small of the back which is connected to the palms of the hands via air filled tubes. Aquanex is a semi-tethered hand pressure measurement system where the swimmer is attached to a boom which is a fixed link to the pressure measurement device on poolside. The swimmer is asked to swim 15m and the pressure exerted by the hands on the water is monitored (Aquanex 2008). Not only is it possible to monitor the pressure exerted by the hands or feet during swimming (Berger 1999; Takagi 2002), it is also possible to monitor the force or pressure exerted on the blocks at the start or on the wall during the turns (Blanksby 1996; Lyttle 1999; ATTRU 2008). Drag on the swimmer through

-3-



the water has also been considered using the MAD system (measurement of active drag) (Strojnik 1999; Toussaint 2002; MAD 2008), which measures the push off force applied by the swimmer during front crawl swimming.

Figure 1 - Monitoring techniques currently used in swimming performance analysis. Physiological monitoring is the fourth main area of interest in swimming performance analysis and is concerned with the continual process of monitoring the overall health of an athlete as well as the outcomes of targeted training sessions. For example the weight, body fat, general health and fitness of an athlete is continually under scrutiny whereas heart rate might only be investigated during heart rate specific training sets where chest strap monitors are used. In addition to active monitoring, where the swimmer is in the water, techniques such as land-based ergometer testing and computer based modelling have been used to model and approximate how the swimmer moves through the water. Computational Fluid Dynamics (CFD)

-4-



using software such as Fluent (Fluent 2008) has been used to model how a swimmer moves through the water. This is particularly successful during the underwater phases of swimming (i.e the glide) but has been less successful during free swimming due to the complexities of the water-air interface (Lyttle 2006). Theoretical models of the hands have also been generated and analysed to calculate the forces produced as the swimmer moves through the water (Berger 1999; Sanders 1999). In addition to modelling the swimmer in the water dry land monitoring using swimming ergometers, such as those manufactured by Vasa (Vasa 2007) and Weba (Weba 2007), which are designed to imitate the actions required for swimming have been carried out. The types of measurements which are possible out of the water are consistent with the types of measurements that are desirable in water, however it is likely that there are differences in the swimming action on the ergometer with respect to in the water. Ergometer testing has been carried out looking into arm and leg power and also the cardiopulmonary response during simulated swimming (Swaine 1996; Swaine 1997; Swaine 2000). These tests are not ideal as they are laboratory based simulations rather than real training however the provision of physiological as well as biomechanical information has the potential of providing a thorough data set which is otherwise not possible to obtain during training.

3- Requirements of an Integrated Performance Monitoring System A research project being undertaken at Loughborough University is concerned with the design and development of an integrated monitoring system which is suitable for use in a swimming training environment. The review of current performance analysis technologies in swimming (above), has led to a detailed requirements specification for an ideal training system. A prototype monitoring system is currently under development with an aim of providing a distributed monitoring system which is as discrete and non-invasive as possible. The system aims to provide both indirect (remote) and direct (continuous) monitoring, which results in instrumentation being fixed onto the swimmer in training and into the pool environment, see Figure 2.

Figure 2 - The identified requirements of an integrated distributed monitoring system. Indirect monitoring takes place remote from the swimmer, within the pool environment, whereas direct monitoring involves equipment being attached to the swimmer in the water. In swimming, video and image processing appears to be the most favoured method of performance analysis as it provides a non-invasive and non-encumbering approach. This type of performance monitoring is an important element within swimming as it is possible for data to be collected during training and competition and the swimmers are able to see themselves on screen. Generally video analysis is used for qualitative discussion rather than quantitative analysis because it is manually intensive and time consuming to provide meaningful quantitative data to the swimmer. For the prototype performance monitoring system the integration and synchronisation of video analysis software is an important aspect of the complete system. This is because video monitoring is such an established and trusted technique which can be used to augment the sensors and devices which are integrated into the complete distributed monitoring system. In addition to image processing and video analysis techniques, force and pressure sensors which are instrumented into the pool surround, such as within the start blocks or timing touchpads, are also required for a fully integrated system. These devices can be used to provide quantitative information to the swimmer regarding the position and time spent on the wall during the turn and the time taken before their first movement off the blocks.

-5-



The third remote monitoring approach is land based physiological monitoring which includes weight, body fat, general health and cardiovascular fitness observations. Continuous monitoring is possible when the swimmer is instrumented with sensor devices and it is envisaged that the distributed monitoring system should include both physiological and physical measurements. This instrumentation should allow inter and intra stroke information about a swimmer’s technique to be monitored as well as differences in technique between individual swimmers. Initial investigations have involved the swimmer using a single tri-axis accelerometer device which logs their motion through the water (Slawson 2008), however inertial measurement units which are capable of logging data for post-processing or transmitting wirelessly to a poolside computer are required to provide continuous quantitative information to the coach and sports scientists based on poolside. Real-time data transmission, processing, analysis and presentation are required to allow the coach to disseminate quantitative feedback to the swimmer during their training session. This will allow the swimmer to make changes in their technique and measure whether these changes are successful in improving their stroke. To ensure that any data collected from the distributed monitoring system provides comprehensive and useful information to the swimmer and coach it is necessary for the data to be linked and archived via a comprehensive web-based training database, using a system oriented approach, to provide a complete performance assessment package (Justham 2008). Data which is collected is required to be collated with any other data from that session and for that swimmer and stored for comparison or retrieval at a later date. In addition data must be analysed and presented in a format which is accessible for the swimmers, coaches and sports scientists to provide in-depth and useful information in a timely manner.

4- Discussion and Conclusions In conclusion the distributed system which is under development at Loughborough University is being designed to provide a novel state of the art performance analysis tool for swimming training to provide timely feedback on all aspects of the swimmer who’s technique is being monitored. This type of tool is critical for the elite swimmer who requires accurate information to develop the most effective training programs. The current research has been focused upon the assessment of existing state of the art technologies and the identification of novel technologies which might be viable within the current application. Existing state of the art technologies provide limited feedback to the swimmer and are restricted by their accuracy, set-up time and obtrusiveness to the athlete. It is envisaged that the system under development will deliver non-invasive, nonencumbering real-time feedback to athletes, coaches and sports scientists in training and competition. In this initial phase, the system is being designed with the elite athlete in mind, however it is envisaged that a successful monitoring system should be capable of being disseminated and used by swimmers of all standards from recreational enthusiasts to the international athlete.

6- Acknowledgements The authors would like to thank the sports scientists, coaches and swimmers at Loughborough University who have taken part in the information gathering stages of this research. Funding for this work has been provided by the Innovative Manufacturing and Construction Research Centre (IMCRC) based at Loughborough University, part of the Engineering and Physical Sciences Research Council of Great Britain (EPSRC), and UK Sport.

7- References [A1] APLabs (2008) "The website of AP Labs which is an Italian research company who have developed the KZ and SPEED systems for swimming: : http://www.aplab.it/homeeng.html, accessed 28th Feb 2008." Volume, DOI: [A2] Aquanex (2008) "Website for the Aquanex pressure measurement system: http://swimmingtechnology.com/Aquanex.htm, accessed 28th Feb 2008." Volume, DOI: [A3] ATTRU (2008). Media press release detailing the features of the AIS Recovery and Swimming Centre in Canberra Australia http://catalogue.ausport.gov.au/fulltext/2006/ascmedia/2006.10.16.asp, accessed 28th Feb 2008. [BH1] Berger, M. A. M., Hollander, A.P., Groot, G.D. (1999). "Determining propulsive force in front crawl swimming: A comparison of two methods. ." Journal of Sports Sciences 17: 97-105. [BG1] Blanksby, B. A., Gathercole, D.G., Marshall, R.N. (1996). "Force plate and video analysis of the tumble turn by agegroup swimmers." Journal of Swimming Research 11: 40-45. [D1] Dartfish (2008) "Manufacturers website of the Dartfish video software solutions company: http://www.dartfish.com/en/index.htm Accessed 27.02.08." Volume, DOI: [DS1] Dekerle, J., Sidney, M., Hespel, J.M., Pelayo, P. (2002). "Validity and Reliability of Critical Speed, Critical Stroke Rate and Anaerobic Capacity in Relation to Front Crawl Swimming Performances. ." International Journal of Sports Medicine 23(93-98). [F1] Fluent (2008). Website for the Fluent CFD modelling software comapny http://www.fluent.com/. This company uses its ANSYS software for modelling the swimmer's motion in the water. Date accessed 27-02-08.

-6-



[G1]

Garmin (2008) "Manufacturers Website for the Garmin GPS fitness products: http://www.garmin.com/garmin/cms/site/uk/ontofitness/ Accessed 27.02.08." Volume, DOI: [JD1] James, D. A., Davey, N., Rice, T. (2004). "An Accelerometer Based Sensor Platform for In-situ Elite Athlete Performance Analysis." Sensors 2004, Proceedings of IEEE 3: 1373-1376. [JS1] Justham, L. M., Slawson, S.E., West, A.A., Conway, P.P. (2008). Business Process Modelling and its use within an Elite Training Environment. International Sports Engineering Association (ISEA) 7th International Conference, Biarritz, France. [K1] Kistler (2008). "Company Website for the Kistler force measurement company: http://www.kistler.com/do.content.gb.engb?content=KistlerCountryHome_KIL Accessed 27.02.08." [L1] Luinge, H. J. (2002). Inertial Sensing of Human Movement, University of Twente. PhD Thesis. [LV1] Luinge, H. J., Veltink, P. H. (2005). "Measuring orientation of human body segments using miniature gyroscopes and accelerometers." Medical and Biological Engineering and Computing 43: 273-282. [LK1] Lyttle, A., Keys, M. (2006). The application of computational fluid dynamics for technique prescription in underwater kicking. Biomechanics and Medicine in Swimming X (Portuguese Journal of Sport Sciences), Porto. [LB1] Lyttle, A. D., Blanksby, B.A., (1999). "Investigating Kinetics in the Freestyle Flip Turn Push-Off." Journal of Applied Biomechanics 15: 142-152. [MJ1] Mackintosh, C., James, D., Grenfell, R., Zhang, K. (2008). Monitoring Sport and Swimming Publish Number: US 2008/0018532 A1. U. S. P. Office. Australia: 18. [M1] MAD (2008). Measurement of Active Drag (MAD) system website: http://web.mac.com/htoussaint/SwimSite/Welcome.html, accessed 28th Feb 2008. [O1] Ohgi, Y. (2002). "Microcomputer-based Acceleration Sensor Device for Sports Biomechanics ~ stroke evaluation using swimmers wrist acceleration." Proceedings of IEEE, Sensors 2002 1(699-704). [OI1] Ohgi, Y., Ichikawa, H., Miyaji, C. (1999). "Characteristics of the forearm acceleration in swimming." Biomechanics and Medicine in Swimming VIII. [Q1] Quintic (2008) "Manufacturers website of Quintic Consultancy Ltd.: http://www.quintic.com/ Accessed 27.02.08." Volume, DOI: [R1] Rootenberg, D. (2006). Inertial and Magnetic Sensing of Human Motion, University of Twente. PhD Thesis. [R2] RSScan (2008) "Manufacturers website for the RS Scan company: http://www.rsscan.com/ Accessed 27.02.08." Volume, DOI: [SH1] Sanders, R., Haake, S.J. (2007). EPSRC Funded research: EP/F006128/1 Improving Swim Performance by Optimising Glide Efficiency and Time of Initiating Post-Glide Actions. 12 Month project running from July 2007. [S1] Sanders, R. H. (1999). "Hydrodynamic Characteristics of a Swimmer’s Hand." Journal of Applied Biomechanics 15: 326. [SJ1] Slawson, S. E., Justham, L.M., West, A.A., Conway, P.P. (2008). Accelerometer Profile Recognition of Swimming Strokes. International Sports Engineering Association (ISEA) 7th International Conference, Biarritz, France. [SB1] Strojnik, V., Bednarik, J., Strmbelj, B. (1999). "Active and passive drag in swimming." Biomechanics and Medicine in Swimming VIII. [S2] Swaine, I. L. (1997). "Cardiopulmonary Responses to Exercise in Swimmer Using a Swim Bench and a Leg Kicking Ergometer " International Journal of Sports Medicine 18: 359-362. [S3] Swaine, I. L. (2000). "Arm and leg power output in swimmers during simulated swimming." Medicine and Science in Sports and Exercise 32(7): 1288-1292. [SZ1] Swaine, I. L., Zanker, C.L. (1996). "The Reproducibility of Cardiopulmonary Responses to Exercise Using a Swim Bench. ." International Journal of Sports Medicine 17: 140-144. [T1] Takagi, H. (2002). Measurement of propulsion by the hand during competitive swimming. The Engineering of Sport 4, Oxford: Blackwell Publishing. [T2] Toussaint, H. B. (2002). "The “Fast-Skin” Body Suit: Hip, hype, but does it reduce drag during front crawl swimming?" Scientific Proceedings – Applied Program – XXth International Symposium on Biomechanics in Sports – Swimming: 15-24. [V1] Vasa (2007). Vasa swim ergometer website: http://www.vasatrainer.com/index.php?page=Swim%20Training%20with%20Vasa%20Trainer%20Vasa%20Ergomet er. Accessed 23/04/2007. [V2] Vicon (2008) "Manufacturers website of the Vicon motion capture company: http://www.vicon.com/applications/life_sciences.html accessed 27.02.08." Volume, DOI: [W1] Weba (2007). Weba Sport Website for their Swim Ergometer: http://www.weba-sport.com/weba/swim_ergo.html. Accessed 23/04/2007. [X1] Xsens (2008) "Manufacturers website of Xsens Motion Technologies: http://www.xsens.com/ Accessed 27.02.08." Volume, DOI:

-7-
