By akademiotoelektronik, 07/01/2023

How artificial intelligence could revolutionize tennis training

Imagine software that would allow a tennis player to guess in advance precisely where his opponent is going to place his next ball. This will soon be possible thanks to artificial intelligence.

Researchers at the University of Queensland, Australia, have developed “machine learning” software, an algorithm capable of predicting the opponent’s next moves with formidable precision. As always for this kind of tool, it is an algorithm that we have trained. He was given thousands of ball strikes from match videos of Nadal, Federer and Djokovic to analyze, but it could be anyone else, coupled with heaps of data on trajectory, speed, angle, foot movements. And by analyzing all that, the style of play, the way a player responds to such and such a shot based on a whole host of parameters, the machine is able to predict not only the type of shot the opponent will hit with an accuracy of 90% but also where the player's next ball will hit the court, less than a meter.

>> READ ALSO - No more French at Roland-Garros: the "black Thursday" of French tennis deciphered in the special episode of "Court n°1"

In other words, this is an extremely valuable indication. A bit like if you had a gift of divination on the intentions of your opponent. Of course, it wouldn't be a question of using that during matches, that would be cheating, but in training, to train specifically against an opponent, to understand how he reacts to this or that type of blow, preparing a strategy would be an extremely valuable weapon. The interest of this tool does not stop there: for the broadcasting of matches, it would make it possible to have smart cameras capable of anticipating the next move and therefore to have perfect image captures.

Virtual matches

How artificial intelligence could revolutionize tennis training

We can also use artificial intelligence to create virtual matches, Federer against Serena Williams for example.

If you wonder what a match between Federer and Serena Williams would look like, or who would win between Nadal and Mc Enroe, or even Federer against himself, you can use a tennis match simulator whose results are quite spectacular . The tool is called “Vid2player”, it was designed by researchers at Stanford University.

Here again, it's artificial intelligence that we fed with video extracts from different players. The machine has learned their style, the way they play and is able to reproduce it in virtual matches. What we see are the digital twins of the players, who imitate their style of play to perfection, their leg movements, their way of hitting on the volley or going up to the net. The level of realism is quite stunning.

You can have fun playing a player against himself or even simulating a French player in the Roland-Garros final, let's be crazy. What's the point? Again beyond the entertaining side, these are tools that could be used in training, for post-match game analysis. What would have happened if you had played such a move rather than another, and the machine will simulate the rest of the exchange to see if it is more effective.

It could also serve to make tennis video games even more realistic in terms of player animation. It should be noted that these are decision-making tools, to improve training, but in sports, there is always a degree of uncertainty. No artificial intelligence will imagine that a player explodes psychologically, for example.

And these very sophisticated tools are nice for the pros, but what about weekend players?

It's becoming more democratic. AI is no longer just for the pros. We are beginning to see the appearance in the general public, for competitive players, of tools, admittedly a little less sophisticated, but very useful, which take the form of simple mobile applications, such as “Swingvision”, created by a former Tesla employee, or “Tennis AI”, which will allow to have an analysis of its performance on the ground.

Just film the game and the software will analyze everything that happens on the field. Where you place the balls, the speed of the strike, the analysis of the placement, give you small summaries of your best shots, advice to improve. A virtual coach thanks to artificial intelligence.

Tags: