Third person view and guidance for more natural motor behaviour in immersive basketball playing

The use of Virtual Reality (VR) in sports training is now widely studied with the perspective to transfer motor skills learned in vir- tual environments (VEs) to real practice. However precision mo- tor tasks that require high accuracy have been rarely studied in the context of VE, especially in Large Screen Image Display (LSID) platforms. An example of such a motor task is the basketball free throw, where the player has to throw a ball in a 46cm wide basket placed at 4.2m away from her. In order to determine the best VE training conditions for this type of skill, we proposed and compared three training paradigms. These training conditions were used to compare the combinations of different user perspectives: first (1PP) and third-person (3PP) perspectives, and the effectiveness of visual guidance. We analysed the performance of eleven amateur sub- jects who performed series of free throws in a real and immersive 1:1 scale environment under the proposed conditions. The results show that ball speed at the moment of the release in 1PP was sig- nificantly lower compared to real world, supporting the hypothe- sis that distance is underestimated in large screen VEs. However ball speed in 3PP condition was more similar to the real condition, especially if combined with guidance feedback. Moreover, when guidance information was proposed, the subjects released the ball at higher - and closer to optimal - position (5-7% higher compared to no-guidance conditions). This type of information contributes to better understand the impact of visual feedback on the motor perfor- mance of users who wish to train motor skills using immersive en- vironments. Moreover, this information can be used by exergames designers who wish to develop coaching systems to transfer motor skills learned in VEs to real practice.