Sports Biomechanics May 2007; 6(2): 215223

Validation of an integrated experimental set-up for kinetic and kinematic three-dimensional analyses in a training environment
A. BOISNOIR1,2, L. DECKER2,3, B. REINE4, & F. NATTA2
des Sciences du Sport et de lEducation Physique, Mont Saint CETAPS UPRES EA 3832 Faculte Aignan Cedex, 2Laboratoire Mouvement Action Performance, Institut National du Sport et de um National dHistoire Naturelle, Laboratoire dAnatomie lEducation Physique, Paris, 3Muse e, Paris, and 4Laboratoire dInformatique Applique e aux Sports, Institut National du Sport et de Compare lEducation Physique, Paris, France
1

Abstract Biomechanical analyses using synchronized tools {electromyography (EMG), motion capture, force sensors, force platform, and digital camera] are classically performed in a laboratory environment that could inuence the performance. We present a system for studying the running sprint start that synchronizes motion capture, EMG, and ground reaction force data. To maximize motion capture (Vicon 612 with six cameras), a special dim environment was created in the stadium. Classical tools were combined with purpose-built tools intended to analyse the different aspects of movement. For example, a synchronization system was built to create a common time-base for all data recordings and a portable EMG system was synchronized by a cable that was disconnected by the athletes movement out of the blocks. This disconnection represented an independent event recorded by different tools. A gap was measured for some sprint start events between kinetic and kinematic (motion capture) data. Calibration results, measurements of time gap, and duration of the independent event were used to validate the accuracy of motion capture and the synchronization system. The results validate the entire experimental set-up and suggest adjustment values for motion capture data. This environment can be used to study other movements and can easily be applied to several sports.

Keywords: Integrated set-up, motion capture, running, sensors, starting blocks, synchronization

Introduction Classically, biomechanical studies propose substitution equipment, such as treadmills and cycle ergometers, to collect kinematic and kinetic data for human movement analyses (Schache et al., 2001b). However, Schache et al. (2001a) noted that ground reaction forces could not be measured for treadmill running trials (except when using treadmills that incorporate force plates, but even then these only record the vertical ground reaction force). This means that the required inverse dynamics cannot be performed to calculate relevant kinetic data. In addition, Schache and colleagues demonstrated signicant differences between overground and treadmill running for all the time distance parameters, such as
des Sciences du Sport et de lEducation Physique, Boulevard Correspondence: A. Boisnoir, CETAPS UPRES EA 3832 Faculte Siegfried, 76821 Mont Saint Aignan Cedex, France. E-mail: alexboisnoir@yahoo.fr ISSN 1476-3141 print/ISSN 1752-6116 online q 2007 Taylor & Francis DOI: 10.1080/14763140701324818

216 A. Boisnoir et al. stride rate and length, and the durations of the stance and swing phases. These results are in line with those of previous studies which found that movement may be modied by an experimental set-up (Wank, Frick, and Schmidtbleicher, 1998). These studies suggested that sporting performances must be studied in natural training conditions. The key difculty when performing biomechanical studies is to obtain the same common time-base for analysis of all variables. The synchronization of measurement tools creates this common time-base for recording the different events of movement data (Rome, 1995). The aim of this paper is to propose the development and validation of a system of synchronized motion capture, electromyography (EMG), and ground reaction force data collection that can be used to study sport movements in a realistic indoor environment. In addition to standard force platform recordings, we also recorded and synchronized the horizontal forces applied to the starting blocks and the instant when the hands are lifting from the starting position. More specically, we arranged classical measurement tools (optoelectronic system, force platform. and EMG system) alongside purpose-built tools nie ` re, 1997; Nouillot and (starting blocks, chronometer, and hand sensors) (Natta and Bre Natta, 2004). Methods Participants Five male high-performance sprinters took part in the study (mean age 21.5 years, s 1.0; stature 1.81 m, s 0.08; body mass 75.3 kg, s 3.8). Twenty-seven skin markers were located on the athletes body, corresponding to anatomical landmarks (Figure 1). Experimental protocol Each trial consisted of a 20-m sprint (time of all the trials: 3.348 ^ 0.097 s) and the sprinter performed four trials with complete recovery between each running sequence. The distance covered by the athletes was around 50 m to a completed deceleration without risk of injury. Data were collected between the start signal and the end of the rst step. Classical measurement tools Optoelectronic motion capture system. A Vicon system 612 consisting of six Vcam cameras was used (200 Hz, 300,000 pixels). In our experimental conditions, the accuracy of the calibration test was 2.872 ^ 0.529 mm. Portable EMG system. Pre-amplied surface electrodes (gain 1000) recorded the activity of seven muscles (trapezius, right pectoralis major, left pectoralis major, erector spinae, gluteus maximus, rectus femoris, and gastrocnemius). Electrodes were placed on the muscle bellies. The EMG of the trapezius, erector spinae, and leg muscles was recorded from the side of the body corresponding to the leg that the sprinter preferred to place on the rear starting block. The EMG data of the gluteus maximus and gastrocnemius served as an indicator of the muscular synergies during the extension and as a comparison with the literature (Guissard and Duchateau, 1990; Mero and Komi, 1990). The EMG portable system (Datalog, Biometrics, France) was xed to the waist of the athletes. It provides eight analog channels and one numerical channel. In this study, seven analog channels served for the muscle recording, the eighth as a reference placed on the wrist. The numerical channel served for the

Integrated set-up in a sprint training environment

217

Figure 1. Body marker placement: M5G and M5D (fth metatarsophalangeal joints), TAG and TAD (heels), MEG and MED (lateral malleolus), CFG and CFD (femoral condyles), GTG and GTD (greater trochanters), ILG and ILD (anteroposterior ilium), 10G and 10D (tenth lateral ribs), ACG and ACD (acromion), VC7 (seventh cerebral vertebra), EYG and EYD (femoral epicondyles), PPG, PAG, PPD and PAD (wrists), TRG and TRD (ears), 1 and 2 (eyes).

EMG synchronization cable. Data were collected at a frequency of 1 kHz on a SD card (128 Mb). Force platform. An AMTI force platform (122 61 cm) was embedded in the track and covered with the same synthetic ooring as the track. Ground reaction force components were studied at the rst step of the sprint start (sampling rate 1 kHz).

Purpose-built tools Starting blocks. Commercial starting blocks were equipped with horizontal force sensors on each block (rear and front), tted with a metallic plate that ensured that the athlete did not place any part of his feet on the track. Thus the sensors captured the whole of the horizontal forces produced by the athlete during the sprint start. Sensors were connected to an amplier (gain 500). Hand contactors. The hand contactors were constructed from a deformed aluminium circular plate. Each produced an electrical signal when the hand was no longer in contact with the ground. Thus it was possible to record separately and precisely the moment when each hand (right and left) was raised. Hand contactors were connected to the same amplier as the starting blocks.

218 A. Boisnoir et al. Chronometer. A digital chronometer (1/1000 s) was started by an electric impulse and was stopped by the breaking of a light beam emitted by a ashlight and captured by a photocell. Experimental set-up Analog data. Two ampliers were connected to the starting blocks, hand contactors, and force platform. A box of analog channels was constructed to supply the six channels from the force platform, the four channels from the starting blocks and hand contactors, and the synchronization channel via the ampliers. All channels were transferred to the workstation through the Vicon patch panel (Figure 2). Synchronization system. For the synchronization system, the start signal (5-V square wave pulse) was activated when the experimenter depressed a foot pedal switch and was simultaneously distributed to ve components: chronometer, light-emitting diode (LED) box, loudspeakers, EMG system, and box of analog channels. A light signal emitted by the LED box was recorded by a digital video camera to control the scene. The electric impulse also triggered the chronometer as well as the loudspeakers emitting the start signal to the athlete. The 5-V square wave was sent from the synchronization system to the box of analog channels and EMG system (Figure 2). Thus the EMG portable system was linked with the

Figure 2. Schematic tool connections. Purpose-built tools: the synchronization box triggered by the pedal emitted the start signal through ve elements: a electric impulse ( ) activated the chronometer, the LED box, and the ) marked the EMG and the analog recording via the EMG synchronization loudspeakers; the positive front ( cable (ESC) and the box of analog channels (box of analog channels), respectively. The EMG synchronization cable was disconnected by the sprinters movement. The starting blocks and the hand contactor, among others, recorded the dynamics of the sprint start. They were connected through an amplier to the box of analog channels. Classical tools: the digital video camera recorded the scene. The EMG system was carried by the athlete and recorded the muscular activity of seven muscles. The force platform was place at the rst step of the sprint and recorded the ground reaction forces. Finally, the Vicon system recorded the rst three steeps of the sprint.

Integrated set-up in a sprint training environment

219

synchronization system by the EMG synchronization cable. This cable was disconnected when it was tightened by the forward movement of the athlete during the start. This allowed us to avoid background noise often present in radio transmission data acquisition. Capture environment. The dimensions of the capture environment were 6 1.5 2 m. The capture environment was equipped with curtains and panels to shield against daylight. This ensured optimal light exposure for the kinematic data acquisition (Figure 3). The extreme conditions of capture required skin markers of larger diameter (25 mm). Data processing Changes in activity in the gluteus maximus and gastrocnemius were analysed from the start signal to the end of the rst step. This analysis was undertaken to compare data with kinetics and previous literature. The accuracy of the measurements was estimated from the calculated value of the centre of mass. With this in mind, a comparison was made between the variation in horizontal velocity of the centre of mass, on the rst step by integrating the acceleration calculated from the force platform signals and the derivative of the centre of mass displacement in the anteroposterior axis. The modelling of the centre of mass was obtained by the application of the model of Zatsiorsky, Seluyanov, and Chugunova (1990) to the skin marker coordinates (Figure 1). To verify the accuracy of data acquisition, we analysed the following variables: . The time interval between the synchronization signal and the disconnection of the EMG synchronization cable (an independent event of sprint parameters). . The stride length from the take-off from the rear block to the rst foot contact. . The four time-events, in chronological order, which were the end of the push-off on the rear block, the end of the push-off on the front block, the rst step landing, and the toe-off. . A gap, which represented a time interval between measurements of the same events collected by the different tools (Vicon, starting blocks, and force platform). For example, the beginning of a step was located on the force platform by a rise of the force curve and was located on the Vicon system by the lower point of the fth metatarsal marker. However, this marker was not necessarily the rst point that contacted the ground. This induced a gap. A paired t-test was used to compare duration and timeevents. Results In the start (push-off duration on blocks), times of horizontal force production on the rear block were 139 ^ 37 ms for the beginning and 316 ^ 34 ms for the end. The rst EMG activity (before force production) was measured in the gluteus maximus muscle at 99 ^ 28 ms. For the gastrocnemius, the activity began with force production on the rear block (138 ^ 23 ms). For both muscles, the end of the activity was measured before the end of force production in the rear block: 227 ^ 34 ms for the gluteus maximus and 274 ^ 30 ms for the gastrocnemius. The force platform data show a rst step, which began at 545 ^ 48 ms and nished at 744 ^ 39 ms. The second EMG activities were measured for the gluteus maximus at 434 ms and for the gastrocnemius at 500 ^ 66 ms. The end of these activities

220 A. Boisnoir et al.

Figure 3. Motion capture environment. HC hand contactors; SB starting blocks; SS synchronization system. Note that curtains and panels allowed the creation of the dim environment necessary for the motion capture. The distance between VCams was 4 to 4.5 m in length and 2 to 7 m in width; the VCams were arranged 3 m off the ground and oriented downwards around the capture volume. The capture volume obtained was 6 1.5 2 m.

was 647 ^ 72 ms for the gluteus maximus and 720 ^ 61 ms for the gastrocnemius (Figure 4). On the rst step, the mean variation in horizontal velocity was 1.18 ^ 0.11 m/s with the force platform data and 1.24 ^ 0.16 m/s with the Vicon system data and after modelling. The statistical analysis (paired t-test) of the kinematics of the centre of mass did not show any signicant difference between the two measurements (t 0.590, P 0.558). During the rst sprint step, the mean variation in velocity of the centre of mass on the antero-posterior axis presented similar values. Disconnection of the EMG synchronization cable produced by the forward movement of the athlete after the start signal generated a negative square wave pulse in the EMG and analog data (force platform, starting blocks, hand contactors, and Vicon system) (Figure 2). The time interval between signal synchronization and this negative square wave pulse provided an indicator for controlling the accuracy of the synchronization on all the measurement tools. The time interval obtained from several trials revealed a signicant linear relationship between analog and EMG data ( y 1.005x 0.003, R 2 0.999). Moreover, the mean difference of this time interval was 0.002 s, but a paired t-test demonstrated that the differences were not signicant (t 1.84, P 0.072). To evaluate whether the optoelectronic data were accurate h, taking into consideration the scale measurements for the sprint, we calculated the stride length from the take-off

Integrated set-up in a sprint training environment

221

Figure 4. Synchronization of kinetic and EMG data. (a) Horizontal force curves: the force of the foot (solid curve) recorded by the sensors of both blocks and the force on the rst step of the same foot (dotted curve) recording by the force platform. (b) EMG data of the gastrocnemius (GA). (c) EMG data of the gluteus maximus (GM). (1) start signal, (2) push-off beginning on rear block, (3) push-off end on rear block (tRBE), (4) rst step landing (tL1), (5) push-off end on front block (tFBE), (6) toe-off (tT1). The EMG data were recorded on the ipsilateral leg.

from the rear block to foot contact (1.33 ^ 0.12 m). As a result, measurement error was estimated at 0.02%. To check the accuracy of time variables, we measured the gap between data collected on various systems (force platform, starting blocks, and Vicon). With the kinetic systems (force platform and starting blocks), the mean values of the four specic time-events of the sprint start are presented in Table I. Results show that the four time-events of both systems (kinetic and kinematic) were signicantly correlated (P , 0.05). The Pearson coefcients of

222 A. Boisnoir et al.

Table I. Event times measured by kinetic systems (starting blocks, hand contactors, and force platform) and gaps between kinetic systems and Vicon system (times in s; mean ^ s). Variables Parameter Event time Gap Paired t-test tRBE 0.316 ^ 0.034 0.017 ^ 0.009 * tFBE 0.494 ^ 0.046 0.019 ^ 0.009 * tL1 0.545 ^ 0.048 0.006 ^ 0.004 * tT1 0.744 ^ 0.039 0.011 ^ 0.005 *

Note: Push-off end on rear block (tRBE), push-off end on front block (tFBE), rst step landing (tL1), toe-off (tT1). * P , 0.05.

correlation were r 0.968 for the end of the push-off on the rear block, r 0.967 for the end of the push-off on the front block, r 0.998 for the rst step landing, and r 0.994 for toeoff. However, the paired t-test showed a signicant difference. Thus, a gap existed in measurements supplied by the two methods when reporting identical events. This gap should be borne in mind during experiments using only one method. The mean values of this gap are presented in Table I. Finally, the mean standard deviation of the gap (13 ^ 9 ms) was of the same order of magnitude as the measurement error (2 ms) between signal synchronization (start signal) and the negative square wave pulse (disconnected EMG synchronization cable).

Discussion and implications In this study, the timing of the leg muscles (Figure 4) corresponded with the literature (Guissard and Duchateau, 1990; Mero and Komi, 1990). These results conrmed the accuracy of the synchronization signal between EMG and kinetic data. The kinematic data of the centre of mass also provided statistically identical results and conrm the precision of measurement. Both of these elements allowed us to validate the use of our experimental set-up. The synchronization of further tools on a cyclic movement allowed us to measure the gap between two different indicators that related to the same event. Results could be used to specify further the occurrence of the same event during the following cycle when other measurement tools are unable to be used. For example, the measurement of the beginning of the rst step was more accurate with the force platform than with the movement of the fth metatarsal marker. Nevertheless, few coaches will have at their disposal an embedded force platform and, even less likely, a succession of force platforms. On the other hand, optoelectronic systems are transportable and could cover a wider range of measurements. Our results validated the entire experimental set-up, which could be used to study motor behaviour. This approach could be applied in many sports activities and could be combined with other measurement tools. Because of the cables that constrain the movement or the background noise with telemetry systems, the use of EMG systems is often difcult for large displacements in a training stadium. The system validated in this study addressed both inconveniences: it allowed the athlete to carry a data acquisition box, which received an external signal before the beginning of the movement without restraining the displacement of the athlete. Many sports activities begin from an initial position before movement and could, therefore, nd this system useful. All results could be integrated by coaches and athletes as feedback. Finally, all the information gathered could have a direct impact on training and performance.

Integrated set-up in a sprint training environment Conclusion

223

When analysing a complex movement, it is often necessary to combine different types of data. Thus, it is essential to synchronize all the systems used for data collection. This exible set-up should be developed according to the objectives of the study. Such an environment is particularly suitable for studying indoor sports. Finally, the interest generated by results obtained in real training conditions points to the need for a permanent experimental set-up.

References
Guissard, N., and Duchateau, J. (1990). Electromyography of the sprint start. Journal of Human Movement Studies, 18, 97106. Mero, A., and Komi, P. V. (1990). Reaction time and electromyographic activity during a sprint start. European Journal of Applied Physiology and Occupational Physiology, 61, 7380. nie ` re, Y. (1997). Effets de la posture initiale dans le de part du sprint chez les athle ` tes fe minines de Natta, F., and Bre haut niveau. Science & Sports, 12, 27s. Nouillot, P., and Natta, F. (2004). Inuence of velocity on the human global postural strategies during the movement leading up to the vertical upside-down position. Neuroscience Letters, 363, 224 228. Rome, L. C. (1995). A device for synchronizing biomechanical data with cine lm. Journal of Biomechanics, 28, 333 338. Schache, A. G., Blanch, P. D., Rath, D. A., Wrigley, T. V., Starr, R., and Bennell, K. L. (2001a). A comparison of overground and treadmill running for measuring the three-dimensional kinematics of the lumbo-pelvichip complex. Clinical Biomechanics, 16, 667680. Schache, A. G., Wrigley, T. V., Blanch, P. D., Starr, R., Rath, D. A., and Bennell, K. L. (2001b). The effect of differing Cardan angle sequences on three dimensional lumbo-pelvic angular kinematics during running. Medicine, Engineering and Physics, 23, 493 501. Wank, V., Frick, U., and Schmidtbleicher, D. (1998). Kinematics and electromyography of lower limb muscles in overground and treadmill running. International Journal of Sports Medicine, 19, 455 461. Zatsiorsky, V., Seluyanov, V., and Chugunova, L. (1990). In vivo body segmental inertial parameters determination using a gamma-scanner method. In N. Berne, and A. Cappozzo (Eds.), Biomechanics of human movement: Application in rehabilitation, sports and ergonomics (pp. 186 202). Worthington: Bertec Corporation.