Soon after the development of the ping-pong robot began, Omron engineers were able to make the robot move the paddle based on the calculations of the sensor and controller.
But no matter how many times it tried, the robot kept missing the ball. It couldn’t even return the ball to the opponent. If a human player is playing with another human player, it is rather easy to maintain a ping-pong rally. However, an extremely high level of technology is required if a rally is to be maintained between a human player and a machine.
The ball a player hits is first detected by the sensor, which then calculates how it should be returned. The controller then controls the robot so as to hit the ball according to the calculated data. This control is extremely precise, with a resolution of 1/1,000 of a second.
Yoshiya Shibata, an engineer in charge of development, recalls that the most difficult part was determining the cause of the robot missing the ball.
“We could not determine whether the problem was in the calculation of the ball hitting position or in the deviation of time from which a hitting command was given. There was also a time lag between the command and the robot’s movement. In addition to these conditions, the ball kept moving all the time. Identifying the cause of the problem in the world of 1/1,000 of a second precision was extremely challenging.”
A machine that can calculate
an easy-to-hit position.
The greatest feature of the ping-pong robot is that it shares a common goal of continuing a ping-pong rally with a human player, and makes calculated judgments about how they can achieve this shared goal.
To make this a reality, it is important that the machine “understands” a person’s conditions and takes action to support the person in a manner suitable for his or her conditions. The ping-pong robot recognizes the opponent’s standing position and paddle position to analyze the position of the ball three-dimensionally and to predict its trajectory. Based on this data, the robot is controlled at high speed and with high precision. Unless these separate operations are synchronized and controlled at 1/1,000 of a second accuracy, the robot ends up missing the ball.
Through this precise control, the robot can return a slow blooper to an opponent who hits a blooper. In the case of an opponent who hits a faster ball, the robot can return a faster ball. Whether the opponent is an adult or child, the robot always returns the ball in a way that is easy for the opponent to hit.
Shibata continues: “Even if a player hits a ball badly, with a trajectory that is impossible for the robot to return, the robot is designed to fully extend its arm in a futile attempt to return the ball. After all, it would be rather unsportsmanlike if the robot simply ignored a ball that was badly hit by its opponent, wouldn’t it?” (laughs)
Omron engineers believe that incorporating this function of cooperation into a machine is essential for the machine to become a truly reliable and helpful partner for people.
We want to deliver benefits of advanced
technology to more people by bringing
machines closer to people.
Let’s take the example of a pitching machine that throws a ball in place of a pitcher.
Many such machines keep throwing the ball even if a batter is tying his or her shoelaces. When we need to re-tie our shoelaces, we must take action to avoid the ball that keeps coming from the machine, like pressing the pause button or moving away from the batter’s box.
In fact, the convenience we enjoy from automation available today still lacks consideration for users and is not cooperative with them.
“But in the future, it will become possible for people to control various kinds of machines simply by telling the machine what they want it to do. We will strive to advance the relationship between people and machines by making difficult logic simple to manipulate,” concludes Shibata.
Omron engineers continue to challenge themselves to create a brighter future for people and machines.
|