This week, new video footage emerged of a four-legged robot goalkeeper making remarkable saves in a laboratory.
Scientists at the University of California, Berkeley, programmed the robotic goaltender, nicknamed “the dog” due to its four legs and canine-like stride.
It is part of an interesting scientific field to create machines that can play sports as well as human professionals, or even better.
FOUR-LEGGED ROBOT GOALKEEPER
Scientists at the University of California Berkeley’s Hybrid Robotics Lab trained the four-legged robotic goaltender, which was demonstrated in a new film this week.
MIT’s Biomimetic Robotics Laboratory developed the four-legged robot Mini Cheetah.
Mini Cheetah lacks a camera, so the position of the ball is determined via an external camera and YOLO, a neural network-based learning technique that provides real-time object detection.
The robot dog was trained using reinforcement learning,’ a form of machine learning that enables artificial intelligence to learn through trial and error utilizing feedback from its activities.
In the video, the dog is subjected to a variety of thrown and kicked shots from humans, as well as shots from A1, another quadruped robot developed by the Chinese company Unitree Robotics.
It can be observed stooping, jumping, sidestepping, and diving to block shots, then returning to its original position after a successful block.
According to experts, it can save 87.5% of shots on goal, compared to the average of approximately 69% for human goalkeepers.
This year, England and Everton’s number one goalkeeper Jordan Pickford had a save percentage of 69.4% across all competitions.
GOOGLE’S PING-PONG ARM
This week, Google also displayed the progress of their ping pong robot arm, which recently competed in a 340-shot match against a human.
More than four minutes after it began, the robot blasted a feeble shot into the net, concluding the historic rally captured on film.
According to Google, the arm plays “cooperatively” to get a decent rally going, although its skills could be improved to match those of a professional human competitor.
Researchers desired to train it in a simulation so that it could interact with humans in the real world but faced a “chicken and egg” dilemma.
How can we replicate human behavior in simulation without having a robot that can interact with a person? stated Google engineer Laura Graesser.
Therefore, the engineers alternated between training in simulation and implementing it in the actual world until they were competent enough to start the rally.
CHINA’S BADMINTON BOT
Researchers in China have constructed the Robomintoner, a robot capable of playing badminton against human opponents.
In 2016, scientists from Chengdu revealed the toy at a competition in Shenzhen, a city in southern China.
According to the robot’s designers, it has a high accuracy rate and a rapid tracking system, allowing it to navigate the court and make respectable returns.
The machine resembles a lawnmower, but it clutches its badminton racket as firmly as any human player.
Using a quick flicking motion, it imparts adequate power to the shuttlecock, and it whizzes around the court on four wheels.
EXPERT SKIER
Experts from the Shanghai Jiao Tong University in China have developed a skiing robot that can regulate its descent down a snowy slope.
The machine stands with two legs on each ski and its middle legs grasp ski poles, which it can manipulate to determine direction.
Experts demonstrated that by putting the machine through its paces on beginner and intermediate slopes, the bot can stay upright, turn, and avoid colliding with people.
The robot was also demonstrated to be capable of skiing at over 30 feet (10 meters) per second over a 1,300-foot (400-meter) slope at an angle of 18 degrees.
Future versions of the robot may be able to compete in robot skiing competitions, patrol mountains, and aid in snow-bound rescues, according to the researchers.
6-FOOT BASKETBALL PROFESSIONAL
The Japanese company Toyota has created a six-foot-tall basketball-playing humanoid with greater precision than professional humans.
According to the developers behind the experiment, the android, simply named Cue, has an accuracy of about 100 percent when shooting from short range.
It learned to score baskets using artificial intelligence and threw over 200,000 practice shots to hone its tactics and accuracy.
In a free-throw shooting competition, the robot scored more free throws than players from Arvalq Tokyo, a team from Japan’s premier professional league.
However, because it is tied to a small platform by electrical cords, it cannot maneuver around the court like a human player.
The design of the android was inspired by the main character of the Japanese manga Slam Dunk, Sakuragi Hanamichi.
CURLY THE CURLING ROBOT
In 2020, the robot ‘Curly’ reportedly defeated one of the world’s greatest curling teams by rapidly responding to changes in the ice.
Curling is a sport in which stones are slid across a sheet of ice towards a circular target, with the objective of landing as close to the center as possible.
Curly tosses the stone, but does not do the task of cleaning loose ice in front of the stone as it goes.
2018’s silver-medal-winning Korean women’s team was defeated in three of four official matches by the robot.
Curly was created by academics at Korea University, who claim that this innovation brings computer simulations closer to the actual world.
FOOTBALL DROIDS
A video from a few years ago demonstrates that an army of football-playing robots might defeat a human team 3-2 in a five-a-side match.
Small droids designed by the Beijing University of Information Science and Technology display incredible speed as they close in on their human opponents.
The droids were trained to avoid allowing the opponent’s time and space on the ball, which is a critical priority for midfielders and defenders.
A mechanism allows the bot to dribble with the ball before a lever flips out, sending the ball into the air toward the goal or to a teammate.
The team was displayed at RoboCup, an international robots competition in which robots compete in matches against one another.
RoboCup was founded in 1997 to build, by 2050, a team of humanoid robots capable of competing with and defeating the best human players.
The soccer champion of ‘THOR’
In 2015, one of the most remarkable winners of the RoboCup was a machine named Thor, which represented the United States.
Thor, a 119-pound, a five-foot-tall humanoid robot constructed by UCLA and the University of Pennsylvania, won the adult-size humanoid robot category with a 5-4 score.
After the competition, the teams gathered for a symposium to discuss the most recent robotics research.
UCLA professor of mechanical and aerospace engineering Dennis Hong described RoboCup as “action-packed.”
‘Thousands of individuals from all over the world with advanced robots and technology are gathered in one location.
Even while it’s fantastic to win, the event is more about friendship than it is about the rivalry.
