Abstract
How does an outfielder catch a ball? The mainstream approaches assume that players’ behavior is controlled heuristically by online information obtained from the optic flow (e.g. LOT, McBeath, et al 1995 or OAC, Chapman 1968). These heuristics do not consider any predictive component and consequently can’t account for unavoidable visuomotor delays. We propose a model that is a generalization of our previous work allowing the observer – at least theoretically – to predict where and when a parabolic flying ball will land. It proposes a prediction that is updated continually as the observer moves as a navigational strategy. We examine how well the model accounts for catching movements collected in one experiment. Participants had to solve a task similar to that of the outfielder problem in an augmented reality setup: We presented them with a soccer ball describing a parabolic movement starting at (x = 0, z = −12) m, with x and z being lateral and depth positions with respect to the observers’ initial position (0, 0). The ball could travel to nine different ending positions that resulted from combining the coordinates x = (−3, 0, 3) and z = (−3, 0, 3) m. The participants used a joystick that allowed them to move in the x-z plane at up to 6 m/s. We showed two different motion durations: 1.5 and 3 s. In both cases, participants used a strategy consistent with the use of a predictive model. Interestingly, when the temporal constraints of the task are more demanding, the predictive strategy produces similar catching movements to those elicited by the heuristics described in the literature. Furthermore, unlike pure online approaches, the interplay of a predictive component allows our model to cope with visuomotor delays.
Acknowledgement: First author is supported by the fellowship FPU17/01248 from Ministerio de Educación (Spain,
https://www.mecd.gob.es) The research group was funded by the Spanish government ref. PSI2017-83493-R (AEI/FEDER, UE).