Feature
No.43
January 2005
 
    

Application of Information Technology to improve Sporting Performance
Arnold Baca & Philipp Kornfeind, Austria
 

Abstract
Information technology provides manifold means to develop sports specific systems, which present feedback information to athletes immediately or shortly after motion execution. In many cases, the system developer may even select between very different approaches to acquire the relevant parameter values to be presented. When designing and constructing systems of that kind, special emphasis should be given to the performance goal. Athletes should be able to recognize deficits. Information should be given, which allows to notice improvements or deteriorations.
Feedback systems have been developed for assisting table tennis players and rowers, a system to be used in biathlon is under development.
In the case of table tennis, tools have been set up to assist players in training. The first system allows to detect the impact position of the ball on the table. Vibration sensors are fixed onto the underside of the table. Positional and timing information is acquired in real time and may be presented shortly after the shot. The second, low cost system, allows to detect and display the times between first and second as well as between second and third impact of the ball on the table. Different service techniques may be trained using this system.
In the case of rowing, a measuring station has been constructed in order to supply coaches and athletes with KP (knowledge of performance)-information based on kinetic parameters. A rowing ergometer is placed onto two force plates. A force transducer is connected to the chain attached at the handle and generates a signal proportional to the athlete’s pulling force. Horizontal (in motion direction) and vertical ground reaction forces and the pulling force are recorded simultaneously and may selectively be displayed in real time, after a short delay or as summary feedback about a series of strokes on a monitor in view of the rower. The application of the system in the training process of elite rowers has shown that subjective observations of the coach could objectively be verified and vividly be presented. Certain deficits in the movement technique could be visualised allowing the athlete to study the effect of changes in the movement pattern.
In the case of biathlon a video based system is under development, which allows to visualize the motion of the barrel of the rifle just before and immediately after shooting.
The success of systems of that kind depends on how exact characteristics to be improved are measured and on how fast and how comprehensible the results can be made available to coaches and athletes

Introduction
Fractions of a percentage point often decide upon success in modern elite sport. Effective ways to improve sport performance are required. Computer scientists and engineers in cooperation with biomechanists, physiologists, sport psychologists and strength and conditioning specialists may interactively develop systems, which provide coaches and athletes with innovative and effective support as they strive for success at the highest level [1].
Time for motor skill acquisition can be reduced by suitable feedback systems. If augmented feedback (Knowledge of Results – KR, Knowledge of Performance – KP) is provided appropriately, sport performance may improve significantly [2]. Consequently, feedback systems have a high value in training. Advances in information technology have made it possible to develop tools and methods, which are specially oriented on the motion task to be performed [3]. Powerful technology and methods are available for developing systems, which allow to provide athletes objectively with supplementary information rapidly and immediately according to Farfel’s principle [4]. Biomechanical and physiological parameter values may be acquired and adequately presented shortly or immediately after motion execution within an effective pre-KR interval [5].
In the design and construction process of a feedback system, special emphasis should be given to the performance goal of the athletes using it. Sports specific limitations and peculiarities should be taken into consideration. The athletes should be able to recognize improvements. It must be evident, what to aim at.
Information technology often provides different methods to assess the problem and to realize a specific feedback system. Pros and cons should thoroughly be analysed before starting the development and implementation. When high performance hardware is used, immediate availability of the results may often be achieved.
In particular powerful data acquisition and analysis as well as video processing tools allow to construct sophisticated sports specific feedback systems. This will be underlined by the following examples from table tennis, rowing and biathlon.

Example 1 – Table Tennis
Factors affecting the quality of the ball played are the spin, the position, where the ball hits the table, and the intervals between the ball impacts on the table. Systems that give immediate optical or acoustic feedback on the quality of the ball just played are applicable in training. Besides of directing and conditioning the technique some motivational effects can be expected.
One exercise, for example, is to return a specific service into a certain area on the opposite side of the table (Figure 1). Another is to play the ball as long as possible. In this case the frequency of an acoustic feedback signal may depend on the distance to the edge of the table. A third exercise is to serve the ball in a way that the interval between the first and second impact of the ball on the table is as short as possible.
Two feedback systems have been developed. The first is based on the detection of the impact position of the ball on the table in real time [6], the second on the acquisition of the ball impact intervals.

Figure 1. Feedback system in table tennis. An acoustic signal is given, if the ball hits the marked area.

Impact position detection
The base system – allowing to detect impact positions on one table half – was developed to give feedback on the accuracy of the placement when performing certain tasks and to give feedback on impact positions and impact times during service. Moreover, it should be possible to evaluate a series of trials and to give summary feedback. For this purposes it was essential to detect impact point coordinates automatically in almost real time.
In service strategy training, however, not only the placement on the opposite half of the table is of interest. The impact positions on both halves of the table and eventually (in the case of short services) that of a third impact point allow additional conclusions to be drawn. The system was therefore designed to be easily extendible to both sides of the table. This (potential) extension is also usable in applications, where the times, when the ball hits the table and the impact positions shall be recorded during a whole match. These data can then be used as input for a match analysis software used in table tennis [7].
Information technology provides different methods to assess the problem. A decision had to been made between methods based on image processing and two different triangulation methods.
Image processing methods would either require the use of a very expensive high speed video camera or that of a (likewise expensive) asynchronous camera taking an image at the very instant when the ball hits the table, thus requiring an additional vibration sensor for triggering. Pre-experiments with video based methods showed that problems might arise from reflections on the table and from the boundary (white) lines on the edges of the table top.
Triangulation methods could either be based on the acquisition and the processing of acoustic signals (sound) or on the acquisition and the processing of mechanical signals (vibrations). The principle is shown in Figure 2. The (acoustic or mechanical) signal originates in the impact point and spreads out. It arrives at the measuring sensors (at least three are required) at different instants of time. From these instants the impact position can be calculated.
To acquire acoustic signals microphones would be applicable. However, due to the high propagation velocity of sound the microphones should allow a very accurate estimation of the instant the signal arrives. Surrounding noise would influence the system in a very negative way.

Figure 2. Principle of the triangulation method. Given the different instants of time (t1, t2, t3) the radii (r1, r2, r3) are determined.
A triangulation method based on the acquisition of vibration signals was realized in the end. Four vibration sensors (one redundant) are fixed on the underside of one half of the table and connected to an amplifier, which itself is connected to a computer based data acquisition system. This system does neither influence the players nor does environmental noise influence the system. It is, however, rather expensive, since very precise vibration sensors (accelerometers; Type: Kistler 8632C10; amplifier 5134A1) and a data acquisiton board recording with a high sampling frequency (National Instruments DAQ 6062; maximum sampling frequency: 500kHz) have to be used. The high sampling frequency required (125 kHz per channel) results from the high velocity of signal propagation through the wooden table (534 m/s; Table: JOOLA Rollomat). Because of different propagation velocities of the signal depending on the material properties of the table top, the system has to be calibrated. An additional problem of the method lies in its dependency on the construction of the table. In the case of the JOOLA Rollomat, a vibration damping material of 6-9 mm thickness is therefore used as an elastic interlayer between the table top and the metallic parts, which are used to stiffen it [6]. This decoupling is necessary due to the fact that vibration signals propagate much faster through the metal than through the wood. Because of that, the signals may arrive earlier at the sensors when taking a way through the metal, even if this is longer. A schematic presentation of the method including the four signals recorded is shown in Figure 3.

Figure 3. Detection of impact points in table tennis using vibration sensors (accelerometers).
The accuracy of the method was tested by drawing a matrix consisting of 165 points (coordinates (in mm) (62+100i, 62+100j); i=0,..14 and j=0,..10; origin of coordinate system: corner of table, coordinate axis in parallel to table edges) and dropping a ball, which was previously pressed into an ink pad, from a height of about 0.3 m onto each of these matrix points (details in [6]). The mean deviation of the centres of the marks and the reconstructed coordinates was 0.020 ± 0.011 m.

Ball impact interval detection
Another system was developed for analysing the quality of service techniques [8]. This (low cost) microcontroller based system (microcontroller PIC16F628) allows to determine and display the time interval between first and second impact of the ball on the table immediately after service execution. In the case of short services, it also determines and displays the time interval between the second and the third impact.
Two microphones are used for recording the signals caused by the ball impact. Both are fixed in metallic boxes. The boxes are put onto both halves of the table. Since intervals between impacts are of interest, it is not essential to determine the times of impact with the same accuracy as it would be necessary when using the data for calculating impact positions. The signals from the microphones are electronically preprocessed [8] and then fed to the microcontroller, which is also connected to the serial port of a PC or laptop. A LabVIEW® 6.1 program acquires the data from the serial port and displays the results (see Figure 4).

Figure 4. Ball impact interval detection during a short table tennis serve. Left: Time between first and second impact on table. Right: Time between second and third impact.
The system works fully automatic. No user intervention is required between successive serves.
Feedback on the quality of different service techniques may be obtained. Typical exercises to be performed by the players include the task to achieve intervals, which are as short as possible, thus reducing the time for the opponent to react properly. It is obvious that this time interval is strongly affected by the degree of spin of the serviced ball.

Example 2 – Rowing
Biomechanical analysis in rowing involves the consideration of the kinematics and kinetics of the boat-rower system. Different authors [9, 10] identify the curve shapes of the horizontal and vertical acceleration of the rower’s centre of mass with regard to the boat and the curve shape of the force applied to the oars as important factors for qualifying and quantifying a rower’s technique. Fluctuations of the boat velocity due to the forward/backward motions of the rower should be kept low, vertical motions of the centre of mass of the rower are uneffective from an energetical point of view, the shape of the force applied to the oars should be bell-shaped.
Analyses of the rowing technique in the boat are difficult to realize and are very demanding in time and instrumentation. In many cases analyses are therefore based on rowing simulators (rowing ergometers) on land.
A feedback system considering the factors mentioned above has been developed for use on land (Figure 5). A rowing ergometer (Concept II®) is placed onto two (Kistler) force plates. A force transducer is connected to the chain attached at the handle. Horizontal (in anterior/posterior direction) and vertical ground reaction force as well as the pulling force are recorded. If the masses of the handle and the sliding seat as well as the vibrations of the ergometer are neglected, the ground reaction forces are directly proportional to the acceleration of the body’s centre of mass of the rower with regard to the environment (the ergometer). Kinematic motion data may be measured simultaneously using a highspeed video system. From a detailed biomechanical analysis of the kinematic data technical deficits may be detected and related to the force data. During successive motion executions on the ergometer (without simultaneous measurement of the kinematic parameters) feedback can be given to the athletes on the quality of their technique. For this purpose, the time histories of the relevant kinetic parameters are displayed on a monitor in view of the rower during motion execution. The rower is thereby able to discover how changes in the movement pattern may alter the shape of the curves in the desired direction. In addition, a series of successive strokes may be evaluated. In this case the results are presented in the form of a summative feedback. A graphical programming language (LabVIEW 6.1) has been used to realize and implement the system.

Figure 5. Feedback system for rowing.
From case studies it has been concluded that the horizontal forces are particularily informative when investigating individual techniques in elite rowing. They show much more inter- and intraindividual differences than those representing the pulling force. As an example, two typical shapes are shown in Figure 6, both from elite rowers. Large differences in the horizontal forces during the pulling phase can be observed (marked area).


Figure 6. Example: Different shapes of horizontal ground reaction forces from two elite rowers.
The system was successfully applied in elite rowing. Subjective observations of the coach could objectively be verified and vividly be presented. To obtain a more mobile system, which is not restricted to the use in combination with force plates, a system, which measures horizontal and vertical reaction forces directly at the foot stretcher (see Figure 5) is under construction.

Example 3 – Biathlon
Coaches and athletes in biathlon are interested in the motion of the barrel of the rifle just before and immediately after shooting. This is a crucial factor because of the preceding high exertions of the athletes. Feedback systems are required, which are able to present this information shortly after having shot.
Again, different strategies may be followed when developing systems capable of these requirements.
On the one hand, a small laser device can be attached to the rifle. By tracking the point caused by the laser beam using position sensitive laser measurement devices, information on the motion of the barrel can be reconstructed.
On the other hand, opto-electronic systems may be used to track the motion of the barrel itself. A method of this kind, based on image processing, is currently under development. The barrel is recorded using a video camera. Tracking algorithms are applied to track the shape of the rifle. Applying the results of a calibration procedure the image coordinates obtained by the tracking method can be converted to object space coordinates. In Figure 7 an image of the barrel is shown. The reconstructed trajectory is shown.


Figure 7. Tracking the barrel in biathlon. The plotted line in the diagram represents the trajectory.
From the sound track recorded by the video camera simultaneously, the instant of shooting may be estimated. First results are promising. It appears that the resolution of video cameras usual in trade (720 × 576 pixel) is sufficient for acquiring the rather small motion of the barrel.

Summary
Real time and rapid feedback systems as well as sophisticated systems for collecting and analysing sports-specific data provide innovative and effective support to coaches and athletes. Powerful IT-tools and wireless technology facilitate the development of user-friendly systems, which are specifically oriented towards their needs. It is of central importance for the success of systems of that kind that the characteristics to be improved are measured exactly and that the results can be made available to coaches and athletes fast and comprehensible. The latter aspect implies that special care has to be taken in the design of the presentation component of the system. A graphical visualisation form should, for example, be preferred to a presentation of pure numbers.
If these aspects are considered, novel and rapid performance measurement and feedback tools based on modern information technology will become more and more pervasive in the training environment.

References
[1]. Broker J. P. & Crawley J. D. (2001). Advanced sport technologies: Enhancing Olympic Performance. In Proc. XIX Int. Symposium on Biomechanics in Sports, J.R. Blackwell (Ed.), University of San Francisco, 323-327.
[2]. Schmidt R. A. & Lee T. (1999). Motor Control and Learning, Champaign, IL: Human Kinetics.
[3]. Liebermann D.G., Katz L., Hughes M. D., Bartlett R. M., McClements J. & Franks I.M. (2002). Advances in the application of information technology to sport performance. Journal of Sports Science, 20, 755-769.
[4]. Farfel W.S. (1977). Bewegungssteuerung im Sport, Berlin: Sportverlag.
[5]. Müller E. (2002). Challenges in optimising human sport performance. In Kinesiology – New Perspectives, Proc. 3rd Int. Scientific Conference, D. Milanovic, F. Prot (Ed.) University of Zagreb, 2-8.
[6]. Baca A. & Kornfeind P. (2004). Real time detection of impact positions in table tennis. In The Enginering of Sport 5, Vol. 1, M. Hubbard, R. D. Mehta, J. M. Pallis (Ed.) Sheffield: ISEA, 508-514.
[7]. Baca A., Baron R., Leser R. & Kain H. (2004). A process oriented approach for match analysis in table tennis. In Science and Racket Sports III , A. Lees, J.-F. Kahn, I. Maynard (Ed.) London: Routledge, 214-219.
[8]. Kornfeind P. (2004). A microcontroller based system for detecting ball impact intervals during a table tennis serve. Internal Technical Report TR-BMCS-01-2004, University of Vienna.
[9]. Nolte V. (1985) Die Effektivität des Ruderschlages. Berlin: Bartels& Wernitz.
[10]. Hill H., Damm K. & Buchholtz C. (1995). Die Optimierung des Kraft-Zeit-Verlaufs des Ruderschlages durch objektives visuelles Feedback. In Rudern lehren, lernen und trainieren, W. Fritsch (Ed.) Wiesbaden: Limpert, 95-104.

Arnold Baca
Section of Biomechanics
Kinesiology and Applied Computer Science
Institute of Sport Science
University of Vienna, Austria
Email: arnold.baca@univie.ac.at



http://www.icsspe.org/portal/bulletin-january2005.htm