"LA BALLE AU BOND : UNE APPROCHE DYNAMIQUE
DES GESTES DE CAPTURE"

Abstract :

Hitting a ball rhythmically with a racket is a perceptual-motor skill that requires the fine control of the timing and force of impact between racket and ball in order to control the amplitude or frequency of the ball's trajectories. In previous work, a model for a ball bouncing on a table was developed in form of a discrete nonlinear map, consisting of an oscillating table and a ball following a parabolic flight trajectory after elastic impact (Schaal, Sternad, & Atkeson, 1996). While a single ball trajectory following one racket contact is entirely determined by the velocity of ball and table at impact, stability analysis of the periodic system revealed that it is the acceleration of the racket at impact that determines the stability of the periodic ball-racket movements.
Specifically, if the racket acceleration at impact is negative and within a given range, the ball's flight is dynamically stable across repeated bouncing. Dynamically stable performance implies that small perturbations need not be corrected explicitly but that they converge back to the periodic attractor. In a series of experiments we demonstrated that human performers hit the ball with negative racket accelerations to produce a consistent bouncing amplitude/period. These results supported the interpretation that coordination of this complex perceptual-motor skill is governed by a single "essential" parameter and humans can exploit the dynamical properties of ball-racket system in terms of a mono-parametric control (Sternad, Duarte, Katsumata, & Schaal, 2000). If humans attune to dynamical stability and racket acceleration at impact is the essential parameter, this parameter should be optimized during the acquisition and improvement of task performance. Three experiments examined the influence of long-term experience on task performance and the change during a short-term learning process. For the experimental apparatus with the specific coefficient of restitution, the range of optimal impact accelerations was calculated to be between -5 and -3m/s2. For all three experiments subjects were instructed to bounce the ball periodically with an invariant ball amplitude throughout the duration of one trial (30sec).
Racket acceleration at impact served as the primary dependent measure. In Experiment 1, six subjects bounced a ball either with their experienced, i.e., dominant, or with their less experienced, i.e., nondominant hand. In accordance with the hypothesis, racket accelerations were more negative (mean: -4.2m/s2) in performance with the dominant hand than with the nondominant hand (mean: -2.0m/s2), implying a higher degree of stability in performance with the dominant hand. These results were associated with less variability in ball ampltiudes in performance with the dominant hand.
In Experiment 2, six subjects performed a sequence of 40 trials of 30sec duration each, using either their dominant or their nondominant hand.
Variability in performance decreased across trials for both hands, testifying improvement. Higher initial variability in the less experienced hand decreased to the same level as the dominant hand. This change was accompanied by an exponential decrease in racket accelerations from positive values to a mean value of -4m/s2. In Experiment 3, six subjects performed the practice sequence of 20 trials first with their dominant hand, followed by 20 trials with their nondominant hand. Another six subjects reversed the order of hand. Again, decrease in variability in the ball trajectory was accompanied by a tendency of racket accelerations becoming more negative in each practice segment. In addition, transfer of skill was observed from one hand to the other. These results support the hypothesis that performance of this perceptual-motor skill involves the exploitation of dynamic stability reducing the necessity for error-based corrections. Skill acquisition involves fine-tuning the essential parameter which ensures dynamically stable solutions of the motor task.
(NSF SBR-97-10312)

Abstract :

Repeated movements can not be reproduced in an identical fashion. Especially in tasks in which subjects try to achieve a constant movement outcome, like in most targeted throwing tasks, this fact is a severe limiting factor for performance. Based on theoretical considerations, that take into account that an unpredictable error occurs during every trial, three different strategies are identified that might contribute to an increase in performance: Stability, Covariation, Noise-Reduction". Data from different throwing tasks will be presented, that demonstrate (i.) that subjects actually use all three strategies, and (ii.) that they increase the extend of that use during acquisition.

Abstract :

Une action d'interception peut être réalisée sans rien connaître des caractéristiques spatio temporelles du point de rencontre avec le mobile.
Pour ce faire, il suffit d'adapter son mouvement en cours d'exécution, sur la base des informations présentes dans les flux perceptifs. Cet exposé est destiné à présenter un certain nombre d'arguments théoriques et empiriques en faveur de cette hypothèse.

Abstract :

Au cours de cette présentation, l'utilisation d'informations relatives à l'accélération du mobile dans les actions d'interception sera questionnée à travers une série d'expérimentations reposant sur des tâches prédictives. Lorsque les participants sont placés devant des trajectoires accélérées ou décélérées présentées dans un ordre aléatoire, la prédiction précise du temps de pré-contact du mobile n'est guerre possible et les participants commettent des erreurs conformes à l'hypothèse d'une utilisation d'information de premier ordre (e.g., tau). En revanche, lorsque la même trajectoire (accélérée ou décélérée) est présentée plusieurs fois de suite, les participants sont capables de minimiser les erreurs. Ce dernier résultat souligne l'existence de stratégies alternatives permettant de compenser l'incapacité du système perceptivo-moteur humain à utiliser des informations rendant compte de l'accélération d'un mobile.

Joe McINTYRE (Laboratoire de Physiologie de la Perception et de l'Action, Collège de France) Titre non communiqué

Contact :

Benoit G. Bardy, Professor
Center for Research in Sport Sciences
University of Paris Sud XI
Batiment 335, 91405 Orsay cedex, France
+33 1 69 15 43 18 (voice) / +33 1 69 15 62 22 (fax)