Core reading
The table tennis robot demonstrates the powerful "behavior" ability of artificial intelligence. In the future, with the upgrading and maturity of table tennis robots, it will help athletes to train better and promote the popularization of table tennis.
Some time ago, a video of "the longest ping-pong round in history" released by the ITTF aroused widespread concern among fans on the Internet. In the video, professional players and table tennis enthusiasts use different tools such as rackets, spatulas and books to relay the ball back in different environments, and staged an interesting table tennis party.
Nowadays, the application of artificial intelligence technology has inserted the wings of imagination into table tennis. A table tennis robot with a smart "brain" is constantly challenging. Will it defeat the table tennis world champion?
Help popularize sports.
In 2014, German table tennis star Bohr faced a table tennis robot named KUKA. In the face of robots, the former world No.1 player took high gear and low gear, but he could only rely on tennis balls and edge balls to keep up with the score and finally win narrowly. Although this duel is only an advertising marketing, the level of robots was not very strong at that time, but the seeds of imagination have been planted in the hearts of researchers.
China Table Tennis Institute of Shanghai Sport University cooperated with Xinsong Robot Automation Co., Ltd. to develop two generations of Pongbot table tennis duet robots. Although the fighting robot has only one mechanical arm, it can already achieve man-machine combat. At the same time, through the deep learning and reinforcement learning model in the artificial intelligence algorithm, the robot can "study" the opponent’s movements and hitting methods while playing, and formulate the return strategy by predicting the table tennis trajectory.
Deng Yaping, the Olympic champion of table tennis, once competed with the second-generation duet robot in a TV program, testing the performance of the robot from three dimensions: speed, strength and rotation. From the beginning of the "helpless" when facing the fast break, to the end, the robot can successfully catch the difficult backspin ball. After one or two hundred rounds of fighting, the robot has shown strong adaptability and learning ability. "Table tennis robots have great potential and cannot be underestimated." Deng Yaping said.
In fact, robots have been widely used in medical care, automobile, chemical industry, education and other fields. In the field of sports, there are also companies that have produced robots that can complete gymnastics movements. Ren Jie, vice president of China Table Tennis Institute of Shanghai Institute of Physical Education, who participated in the research and development of table tennis robots, said that the original intention of designing table tennis robots was to use scientific and technological means to help athletes train better and promote the popularization of table tennis.
Strong learning ability
In March of this year, Pombert played against the robot’s "brother" product — — Serve robot comes out. The service robot has two mechanical arms, which can throw the ball with one hand and hit the ball with the racket in the other. Compared with the traditional service machine, the service robot highly restores the service action of real people, provides athletes with time to observe their opponents, and exercises their reaction ability and ball handling ability.
However, in order to achieve a high degree of personification, in addition to the appearance and action, we must also realize the variety of service. Ren Jie introduced that the service robot can set different rotation types and landing points of the ball, and can simulate different playing styles through the quick change of the racket. "After practicing such a combination, I feel as if I am playing against a real person, or a master!" An experiencer lamented.
The "trump card" of the serve robot is the function of trajectory analysis and action analysis. Two distributed binocular vision systems are like a pair of "smart eyes", which can capture the trajectory of the ball and the athletes in time. Coupled with the "brain" of artificial intelligence algorithm for deep learning and timely adjustment of technical and tactical strategies, a set of "combination boxing" makes the service robot "smart".
"If you beat the human Go master ‘ Alpha dog ’ Show us the powerful artificial intelligence ‘ Thinking ’ Ability, the table tennis robot shows the powerful artificial intelligence ‘ Behavior ’ The ability reflects the manufacturing strength in the field of industrial automation in terms of accurate identification, rapid response and flexible control. " Ren Jie said.
With a deep understanding of table tennis, China Table Tennis Institute has avoided many detours in the process of cooperative research and development of table tennis robots. At the beginning of the design, the serve robot was set at a very fast speed. Later, it was found that this exceeded the acceptance ability of most athletes, which not only failed to achieve the exercise effect, but also increased the configuration of motors and motors, which was completely unnecessary. "Technology research and development should conform to the laws of reality." Ren Jie said, "So we suggest debugging the robot with reference to the speed, frequency and rotation of the ball when the coach conducts multi-ball training."
Wide application prospect
In the past table tennis class, 20 students shared 10 tables. If the teacher practiced multi-ball training one by one, it was often difficult to take care of all the students in one class. Gu Nan, director of the training department of China Table Tennis Academy, said that after the school starts, the college will put 10 service robots into classroom teaching. Through personalized settings, each student can carry out targeted training in a mode suitable for his current level. In this way, teachers have more time to guide students.
Robots can assist teaching, can the training object be robots? China Table Tennis Academy makes serve robots and duet robots become "apprentices". The former uses professional training mode and flexibly adjusts course training according to the progress level of apprentices. The latter quickly learns different return strategies.
If the goal of the duet robot is to beat the world champion of table tennis, then the service robot takes the route of "being close to the people", which is more suitable for ordinary table tennis lovers to practice basic skills. In the future, in addition to daily teaching, service robots can also enter community fitness centers, table tennis clubs and so on. With such a machine, one person can also play ball. If you set the relevant data, you can have a hearty duel.
Ren Jie imagined that in the future, with the help of technologies such as 5G and AR (augmented reality), robots may also be able to synchronize the movements of the participants at the other end of the line in real time, realizing the "cloud battle" of table tennis matches, making table tennis break through the limitations of time and space and become more interesting. Shi Zhihao, dean of China Table Tennis Institute of Shanghai Sport University, hopes to lead the development trend of table tennis in the world through the combination of artificial intelligence technology and table tennis. At the same time, in the process of upgrading the related products of table tennis robots, more technological innovations will be applied to the cultivation of high-level table tennis talents.