Dog-Like DribbleBot Is the Lionel Messi of Soccer-Playing Robots

Soccer is that European kind of football where two teams of 11 players chase after a ball and kick it as they try to drive into the goal. Whereas the good-old American football relies on team strength and tactics, soccer is more of a skill sport, with talented individuals capable of turning all that ball kicking into a ballet of sorts.

One of the sport’s most revered names is Argentinian Lionel Messi. The guy had a tremendous 2022, winning everything that could be won, including the World Cup, and currently working his magic for French club team Paris Saint-Germain.

But Leo Messi has one big fault: he’s human, hence he can’t be of any use to a robot soccer league, in case something like that ever gets here. And it just might, if we are to judge by the statement in the first paragraph, and the DribbleBot, which may just become the virtuoso of its species.

The thing is the creation of the Massachusetts Institute of Technology’s (MIT) Improbable Artificial Intelligence Lab. Like many other robots we keep seeing recently, it’s of the headless dog-like variety, only unlike the others it’s not meant to perform a specific, useful task, but built to play soccer.

Sure, like all robots of this kind, it can move over sand, gravel, mud, and snow, and even climb stairs, but it can do all that while handling a ball. Don’t believe me? Have a look at the video attached below this text, and a new level of panic should engulf you.

Now, to be fair, despite its name the robot does not actually dribble opponents, and its skills are limited to just driving the ball around. Even so, given how we’re talking about the first robot capable of doing this, other of his species might just see it as the Messi they needed.

For MIT researchers, the DribbleBot is not about teaching machines to play sports. They used this approach as a means to learn how to actuate robot legs during dribbling, as that may help such things better navigate the various environments of our world.

That’s because, unlike simple walking, dribbling requires the robot to adapt locomotion as to pursue its target, it has to react to the way in which the ball is affected by the surface it travels over, and so on.

We don't know when this research will become useful for real-life applications, but considering DARPA, the U.S. Air Force Research Laboratory, and the U.S. Air Force Artificial Intelligence Accelerator are all involved, we can almost bet this is going somewhere.

For now, we’ll have to settle for a paper on the DribbleBot, to be presented in full at the 2023 IEEE International Conference on Robotics and Automation (ICRA) at the end of May 2023.