Towards the application of virtual reality training to improve the performance in sports by the example of karate

 
Methods

For both studies, we created virtual characters using motion capturing on the basis of five high-skilled athletes who performed several attacks which were chosen based on previous competition analyses. Afterwards, we carried out two training studies.

Study 1: training of response behaviour and attack recognition using HMD

Fifteen experienced karate athletes responded to sports specifically in a VR training of ten sessions (ten minutes each) over six weeks to attacks of a virtual nature which they saw in a HMD (Fig.1, A). The VR training was integrated into the course of the regular physical karate training. Before and after the intervention, we analysed the following parameters: time to response (time between the attack initiation of the virtual opponent and the first response of the reacting athlete); response quality (measured by a score system depending on whether the athlete could respond adequately); kind of response (if the athletes performed an evasive movement or a direct and successful attack); and attack recognition. The last mentioned parameter was examined using different reaction times based on literature values which were subtracted from the first response of the reaction athlete in the video material to analyse to which movement phase the responding athlete actually reacted in the attack. A further control group received only the regular physical karate training.

Study 2: cognitive training using a VR app

In this VR study (Fig.1, B), six karate athletes watched movement executions of a virtual character in eight sessions (again ten minutes each) over two weeks in a seated position in an application for smartphones which they latched into a HMD. Furthermore, detailed information concerning the correct movement was given in writing. In the VR app, the athletes could self-reliantly choose the perspective and the speed. Another six athletes trained with a video which they saw on their smartphones and a further control group of six athletes received no additional cognitive training.  Again, a video analysis was carried out to examine own movement executions pre-test and post-test of the eighteen karate athletes on the basis of biomechanic parameters which were determined with an international successful karate coach.

Results

Study 1

Using Friedman-tests with subsequent Bonferroni post-hoc tests, the estimation of effect sizes and students’ t-tests, we could find significant improvements in time for response, response quality and kind of response for the VR training group, but not for the control group. Moreover, attack recognition improved significantly. Due to the VR training, the athletes could respond earlier and better, and were able to recognise attacks in very early movement stages, during stepping and shortening of distance between both athletes during the attack preparation phase. For further detail see Petri et al. (2019a,b).

Study 2

By use of ANOVAs and t-tests, we could show that the VR group could significantly improve their own movement executions compared to the video group and the control group.

Based on the results of both studies, we conclude that VR can be used for physical and cognitive training to improve response behaviour and own movement execution as well as to deepen knowledge about karate techniques.

Future directions

We assume that VR training with a session length of 10 minutes each is credible regarding symptoms of cybersickness. The athletes had fun fighting against the virtual opponent and preferred the VR app over the video due to the greater degree of realism and involvement. Therefore, that such VR training improves the performance may serve as motivation. Independent cognitive training with the VR app is useful to deepen knowledge while it does not lead to overload. Future studies should increase the sample size and also include a transfer test to analyse whether the positive effects found in study 1 are also visible in reality.

Acknowledgements

The work was supported by the German Research Foundation under grant WI 1456/17-1. We thank for the good cooperation with the Fraunhofer Institute for Factory Operation and Automation IFF, Germany. Furthermore, we thank Prof. Dr. Claus-Dieter Ohl, Dr. Peter Emmermacher and Christian Timmerevers for their support in the creation of the VR app.