Man vs Machine: Google’s ping pong robot rally of 340 hits – and the tech giant says it could soon take on the PROS
- Google engineers posted a four-minute video clip of the monumental match-up
- Bot’s skills are developed with realistic and simulation-based training scenarios
- The long rally ends when Google’s creation sends a weak shot in the net
New footage shows Google’s ping-pong robot in an epic 340-shot rally with a human.
The advanced robotic arm glides effortlessly along its side of the ping pong table while answering human shots.
The monumental rally ends when the robot sends a weak shot into the net more than four minutes after it started.
Google used realistic and simulation-based training scenarios to turn the robot into a competent ping-pong player.
It currently “plays” with a human to get a good rally going, but according to Google, its skills could be built up to match any human’s level.
The robot, dubbed i-S2R, glides effortlessly down its side of the ping-pong table while returning the human’s shots
WHAT IS i-SIM2REAL?
Google says the robot uses a training platform called i-Sim2Real, short for “real life iterative simulation.”
In i-Sim2Real, knowledge gained by the robot during a simulation is used in a real environment.
i-S2R also alternates between training in simulation and deployment in the real world to build its skills.
The robot uses a training platform called i-Sim2Real, which stands for ‘simulation to real life’.
Researchers had wanted to train it in a simulation so it could play with people in the real world, but they had a chicken-and-egg problem.
‘How do we collect examples of a human interacting with a physical robot to simulate human behavior without already having a robot that can communicate with a human?’ said Google engineer Laura Graesser.
Engineers therefore alternated between training in simulation and deployment in the real world until it was good enough.
‘In each iteration, the human behavioral model and policy are refined,’ says Graesser. “We’ll stop repeating when we can finally have fun playing with the robot.”
Google says future work would build a robot “coach” that could adapt its playstyle to the skill level of the human participant to keep things “challenging and exciting.”
However, engineers admit that table tennis is a “fast, dynamic task” that requires the two players to react quickly to each other’s movements, making it a “challenging test bed for human-robot interaction research.”
The monumental rally ends when the robot sends a weak shot into the net more than four minutes after it started
In the video, the human engineer generally sends shots through the center of the table, though Google says it can “match the precision of amateur humans” in its return and placement.
Both man and machine use a right-handed forehand orientation all the time, but despite this, the bot shows some impressive reaction speeds.
According to Google, i-S2R can also play table tennis against a human or another robot.
Google describes the technology as “a research platform for agile robotics” that can answer questions about human-robot interaction and collaboration.
Google said it used realistic and simulation-based training scenarios to build the robot’s skills
“Robot learning has been applied to a wide variety of challenging real-world tasks, including dexterous manipulation, leg movement and grasping,” Google researchers say in a statement. blog post.
“It’s less common for robotic learning to be applied to high-acceleration dynamic tasks that require close human-robot interactions, such as table tennis.
‘In this project, the goal of the robot is cooperative in nature: to maintain a rally with a human for as long as possible.’
This isn’t the first table tennis-playing bot; Japanese robotics company Omron has previously shown its table tennis bot, called Forpheus, at CES in Las Vegas.
Forpheus has sensors that collect data about the opponent’s facial expressions and physical movements to play a match.
However, not all sporting ventures involving robots have gone smoothly; Earlier this year, a chess robot broke a child’s finger during a tournament in Russia.
CHESS ROBOT BREAKS SEVEN-YEAR-OLD BOY’S FINGER DURING TOURNAMENT IN RUSSIA
A chess-playing robot broke a child’s finger during a tournament in Russia in July – and the incident was captured by CCTV.
The robot grabbed the finger of the seven-year-old boy because he was confused by his overly fast movements, Russian media reported, citing the president of the Moscow Chess Federation – who apparently blamed the child.
“The robot broke the child’s finger – this is, of course, bad,” Sergey Lazarev told the Russian news agency TASS, keeping his organization at a distance from the robot.
The incident took place during the Moscow Open on July 19. Lazarev said the federation had rented the robot for the event, which ran from July 13 to 21.
Lazarev said the machine had been rented for many previous events without incident, and said the boy moved a lot too fast after making a move.