Humanoid robots now play tennis like Nadal and Federer

Researchers have successfully taught humanoid robots the movements and playing style of some of the world's top tennis players using images from professional tennis matches.

In this project, the movements of players like Rafael Nadal, Roger Federer, and Novak Djokovic were extracted from television videos of major tournaments and then used for simulation on the robots. This technology allows the robots to not only perform basic strokes like serves and rallies but also to mimic the unique movement characteristics of each player.

This system, named AdaPT, was developed in collaboration with Novitum, Dobot, and Shanghai Jiao Tong University. The goal of this technology is to transfer the athletic skills of professional humans to robots through the analysis of movement data.

One of the main challenges in this process was the difference in the robot's performance in a simulated environment compared to real conditions. To increase accuracy, researchers designed a method called "dynamic speed adjustment" that helps the robot adjust its movement speed based on the speed and trajectory of the ball.

The AdaPT technology has been tested on two humanoid robot models, namely Unitree G1 and Dobot Atom. These robots do not use a camera on their body to detect the position of the ball and their movement; instead, motion data is recorded by an external tracking system equipped with 35 cameras in the laboratory environment.

This achievement could pave the way for a new approach to teaching complex human skills to robots in various fields, from sports to daily activities.

Source: Zoomit