Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124

Before Atlas the humanoid robot entered the field to hand the ball to the referee during the Norway and Brazil matches World Cup In Sunday’s game, he flashed his football skills on the sidelines.
At the end of the first half, Atlas emerged from the players’ tunnel and repeated a series of distinctive goal celebrations before passing the ball. But it seems that the robot was shy, because it is actually capable of much more than that.
In a series of Videos posted on YouTubeBoston Dynamics shows how it trained a humanoid robot to perform a number of soccer tricks, including its own version of the rabona — a complex move in which the kicking leg crosses behind the standing leg to hit the ball — which the company calls the rabona ghost.
when I met the latest version of Atlas At CES last January, I had no idea that by the summer it would be able to pull off World Cup-worthy moves. But I shouldn’t underestimate it – after all, this robot, and many like it, are designed to constantly learn new things.
These humanoid robots will be deployed first in industry before moving into service and entertainment settings, and eventually into our homes. That’s still a long way off, but the learning they do along the way is crucial to getting there.
In the meantime, it’s important for Boston Dynamics to share Atlas’ skills with the world — and not just for entertainment purposes, says the company’s director of robotic behavior, Alberto Rodriguez.
“It’s a public service to show that technology is reaching a certain level of capability,” he says.
Not only does it spark debate about how this technology fits into society, but it also raises public awareness of how close we are to humanoid robots becoming commonplace.
I’m curious as to why, of all the things Atlas could learn, Boston Dynamics would want to teach a robot soccer skills.
“We’ve always been inspired by high-strength physical behavior or agility,” Rodriguez says. “It motivates us to achieve more of the performance that we know is possible with the robots we build.”
Training for Atlas to be ready for the World Cup began by using motion capture to record the movements and skills Boston Dynamics wanted the robot to perform. They were then placed in a simulation, and “through tremendous trial and error,” Atlas then learned to imitate these movements as best he could within his physical limitations, Rodriguez explains.
He adds that there are two levels of robot mastery of skills. The first part of this relates to the robot’s limbic system – balance and balance, agility and movement. He needed to develop lightning-fast muscle memory, which he also needed for athletic performances in the areas of dance or gymnastics.
The second level was more difficult, going beyond the athletic level. It involved the robot’s manipulation of objects and its ability to exert just the right amount of force to manipulate the world around it.
Teaching Atlas how to automatically adapt to friction and slip, as well as being precise in how close it got to the ball, really pushed the robot outside of its comfort zone. Rodriguez says the model was much more difficult to design than, say, a backflip. “It’s all in the air, where the dynamics are well understood and easy to represent in simulation.”
Atlas may not boast an exact replica of human physiology, but it has been designed in such a way that it is able to mimic human “fluidity and dynamism.” But that doesn’t mean football education has been without growing pains.
In the Boston Dynamics Soccer School video series, it’s clear that Atlas had a whole host of missteps on his way to mastering the skills. Rodriguez says teaching Atlas athletic skills is particularly challenging because the process inevitably involves contorting his body into positions that put him at risk for a “catastrophic fall.”
However, breaking and repairing is part of robot training, he adds, and there is a “good process” for fixing them. And by the time we see them – walking out onto the football field, for example – we are unlikely to see them falling.
“When we deploy robots, they tend to do things that are already well-tested, and we are confident that they will not get into critical situations,” Rodriguez says.
Atlas is already more skilled than a lot of less athletic, creaky humans when it comes to soccer, but I asked Rodriguez if there were any skills he wished Atlas had learned that he couldn’t teach the robot by the World Cup.
“It’s not hard to learn to kick a ball, and we certainly did,” he says. “But kicking it really well, it’s really hard to do.” He pointed to the way legendary footballers such as David Beckham and Roberto Carlos were able to dramatically bend the ball towards their intended goals.
“This is the thing that you eventually probably have to learn through practice in the real world,” he says. “This is a very difficult thing to learn in a simulation.” “Maybe you have to learn through practice and error using a real soccer ball.”
Will Atlas join the team in 2030?
Fortunately, Atlas has four more years to master this skill before the next World Cup. Should we expect that by the time the 2030 tournament rolls around, Atlas may have been hired by a team?
Despite rapidly developing soccer skills, it’s unlikely we’ll see humanoid robots playing on humanoid robot teams, Rodriguez says. What’s more likely is to see teams of robots playing against each other.
Robots can move in ways that human players cannot, such as rotating their joints or flipping their limbs, allowing them to turn without having to take any steps, for example. This won’t make them better players, but it will undoubtedly change the way the game is played in a way that might be difficult for a mixed group of bots and humans to navigate.
Meanwhile, Atlas has learned a tremendous amount from its foray into football. Its new movement, accuracy and speed may not get it a World Cup trophy anytime soon, but overall the robot has leveled up.
“Forcing ourselves to follow these behaviors has had an indirect effect in improving the way Atlas works overall,” Rodriguez says.