Our world, tomorrow by Viviane Lalande / Scilabus

Using virtual reality to boost skills in Olympic boxers

Franck Multon, Inria Research Director at Université de Rennes and Alan Wagner, Assistant professor in the Department of Aerospace Engineering and research associate in the Rock Ethics Institute
On October 4th, 2022 |
3 min reading time
Franck Multon
Inria Research Director at Université de Rennes
Key takeaways
  • The REVEA project aims to use virtual reality (VR) for sports training.
  • With the 2024 Olympics fast approaching, this project will help athletes improve certain “sub-skills” such as speed and motor coordination.
  • To overcome the limitations of a real training session, a VR simulator has been developed based on measurements of athletes' movements in 3D.
  • The project is part of a more general sports programme in which physical, mental, and technical preparation complement the VR sessions.
  • The system is constantly being improved through feedback from coaches and athletes, but also by applying new AI research.

The 2024 Olympic Games are just around the cor­ner and it’s not just the city of Paris that’s get­ting ready… sci­ence is get­ting involved too! At Inria Rennes, we are con­duct­ing research with the Olympic box­ing team and its coach: as part of the REVEA project1, we are look­ing at how vir­tu­al real­i­ty can help the team work on cer­tain “sub-skills” that need to be improved, such as speed, motor coor­di­na­tion and strength train­ing, to name a few. Con­trary to what one might expect, our vir­tu­al real­i­ty sys­tem does not aim to repli­cate a real match or com­pe­ti­tion, but rather to iden­ti­fy a spe­cif­ic sub-skill and improve it in a tar­get­ed com­bat sit­u­a­tion. In the case of box­ing, the sub-skill we are look­ing to train is the abil­i­ty to antic­i­pate oppo­nents’ attacks.

Cred­it : REVEA.

To analyse the oppo­nen­t’s behav­iour with­out vir­tu­al real­i­ty, two box­ers must com­bat face to face, with one of them being asked to repro­duce typ­i­cal attacks. It is very dif­fi­cult, how­ev­er, to con­trol and faith­ful­ly repro­duce cer­tain move­ments from one tri­al to the next. This approach inevitably lim­its train­ing time since such an exer­cise is phys­i­cal­ly demand­ing when real punch­es are being exchanged. In vir­tu­al real­i­ty, this type of train­ing can be done with our avatar and a vir­tu­al oppo­nent with­out the risk of any­one being injured. It also offers a range of stim­uli that go beyond what is pos­si­ble in real life: we can exper­i­ment with var­i­ous visu­al and audi­to­ry per­cep­tions or prac­tise with a high­ly-skilled oppo­nent to push the lim­its of tra­di­tion­al training.

A boxer “behaviour map” 

To this end, our team has devel­oped a vir­tu­al real­i­ty sim­u­la­tor that uses three-dimen­sion­al mea­sure­ments of the move­ments of pro­fes­sion­al box­ers obtained with motion cap­ture devices – via the Immer­move plat­form installed in the M2S lab­o­ra­to­ry) The data obtained from this sim­u­la­tor, which includes, for exam­ple, antic­i­pa­to­ry defen­sive move­ments, such as slight hip move­ments, is then trans­posed onto vir­tu­al humans with dif­fer­ent mor­pholo­gies. As such, the box­er can visu­alise his oppo­nent, rep­re­sent­ed by an avatar and dis­played in a vir­tu­al real­i­ty head­set. In our exper­i­ments, we observe the box­er’s reac­tions to dif­fer­ent types of blows, and whether he is sen­si­tive to the slight antic­i­pa­to­ry move­ments per­ceived by the oppo­nent. These obser­va­tions can be quan­ti­fied, for exam­ple, by suc­cess rates or reac­tion times, to cre­ate a “map” of the box­er’s antic­i­pa­to­ry performance.

As well as apply­ing it to box­ing, we used our approach to train goal­keep­ers at the Stade Ren­nais Foot­ball Club. Our vir­tu­al real­i­ty sys­tem trained them not to focus on the play­er with the ball, but to bet­ter man­age periph­er­al vision infor­ma­tion. Indeed, if the foot­baller about to attempt a goal makes a pass to some­one on the oth­er side of the goal at the very last sec­ond, the goal­keep­er may not have time to react.

Applied research

Our research is very applied: our objec­tive is to help the Olympic teams get ready for Paris 2024. Indeed, think­ing about how to include these tech­nolo­gies in the dai­ly life of ath­letes is one of the main goals. We asked our­selves the fol­low­ing ques­tions: how much train­ing should be done with this type of tech­nol­o­gy? At what time of the day and at what stage in the train­ing pro­gramme? How long should a typ­i­cal ses­sion last? Who should we train with – Olympic hope­fuls or all ath­letes, for exam­ple? And how often? Our approach is part of a wider train­ing pro­gramme set by the coach, which includes phys­i­cal, men­tal and tech­ni­cal preparation… 

The train­ing pro­gramme includes phys­i­cal, men­tal, and tech­ni­cal prepa­ra­tion… with added vir­tu­al real­i­ty sessions!

Our sys­tem is being used at the Olympic team’s train­ing cen­tre and by the sci­en­tists at INSEP work­ing with the French Box­ing Fed­er­a­tion. We are con­tin­u­ing to devel­op it and make it more effec­tive thanks to the feed­back we receive from ath­letes and coach­es. For exam­ple, in our first ver­sion, the oppo­nent made too many repet­i­tive move­ments, which made the com­bat unre­al­is­tic. We are also improv­ing the equip­ment used for vir­tu­al real­i­ty training. 

“Translating” the data

We have a soft­ware plat­form that is based on a com­mer­cial tool called Uni­ty, rou­tine­ly used for mak­ing video games. It loads a three-dimen­sion­al envi­ron­ment – in our case, a box­ing ring – as well as vir­tu­al humans and their move­ments. We write a sce­nario in which the move­ments of the vir­tu­al humans are stored in “move­ment libraries”, which we play back. There are then algo­rithms linked to the motion con­trol of the avatars. All these ele­ments pro­vide us with the met­rics that allow us to mea­sure the reac­tion times and the antic­i­pa­to­ry behav­iour of the boxers.

We have also start­ed to work on arti­fi­cial intel­li­gence, neur­al net­works and deep rein­force­ment learn­ing to improve the behav­iour of vir­tu­al oppo­nents. This part of the tech­nol­o­gy is not yet mature, but ini­tial results show that it is pos­si­ble to repro­duce the com­plex inter­ac­tions between two box­ers with very few obser­va­tions of real com­bats. Tak­ing this approach to its lim­its, we could con­sid­er sim­u­lat­ing the move­ments and strate­gies of a par­tic­u­lar box­er based on some video footage of his pre­vi­ous matches.

Interview by Isabelle Dumé 
1The REVEA project is financed by the pri­or­i­ty research pro­gramme (PPR) « Sport de très haute per­for­mance », part of the Pro­gramme d’In­vestisse­ment d’Avenir, set up in the run-up to the Paris 2024 Olympic and Par­a­lympic Games. The pro­gramme is co-spon­sored by the Min­istry of High­er Edu­ca­tion, Research and Inno­va­tion and the Min­istry of Nation­al Edu­ca­tion, Youth and Sport.

Our world explained with science. Every week, in your inbox.

Get the newsletter