The neural network has learned to simulate ultra-realistic tennis matches

American developers have created a neural network that simulates tennis tournaments and predicts the actions of real players. It can be used to analyze past matches and in the process of training.

How it works: The neural network was trained using videos of matches and training sessions. Thanks to the cyclical nature of tennis tournaments, experts have been able to develop a statistical model that predicts how real tennis stars will behave in specific situations. Artificial intelligence takes into account not only the overall strategy of the game, but also the individual style of each player. For example, the features of the ball, favorite strikes and tactics. This is what distinguishes the development from existing systems that can simulate the game of tennis.

In addition, the new neural network can simulate matches that will not take place in real life. For example, when a tennis player plays against himself or an opponent of the opposite sex. It also allows you to set the position of the player during the serve and change the place where the ball lands.

Why is this needed: In addition to an attempt to brighten up the frustration of tennis fans from the cancellation of major championships (and the inability to buy tickets for those that will still take place, but without spectators), the technology can be used in the process of training athletes and for analyzing past competitions.

What a neural network cannot: the only drawback of artificial intelligence is weak visuals. So far, the model is more reminiscent of video games from the nineties: the spectators in the stands are static, the objects do not cast shadows, and the players “glide” along the surface of the court. In the future, the specialists intend to work out the graphics and thereby improve the visual component of the neural network. Otherwise, it is correct and can already be used for its intended purpose.

Who did it: a team of scientists from Stanford University. They decided to create such a neural network due to the cancellation of the annual Wimbledon tournament due to the coronavirus pandemic.

Bonus: a segment of the video where Roger Federer plays with himself (from 0:48):


Subscribe to the Trends Telegram channel and stay up to date with current trends and forecasts about the future of technology, economics, education and innovation.

Leave a Reply