In fact, many things were developed very early, but the approaches were of course different, because the computing power available at that time was not the same. The system had two VME processor boards, each with a Motorola 68020 processor with 25 HMz. One was responsible for calculating the control algorithms and the other for calculating the game strategy. The image processing was solved purely electronically. The interlaced signal contained two fields of a video frame captured consecutively with a full frame rate of 25 Hz. Each frame were evaluated, resulting in a frame rate of 50 Hz. A binary image was generated from the camera signals and the centre of gravity of the white pixels was determined in the image. In order to suppress interferences, a window could be defined in which the center of gravity determination was applied. So as soon as we knew where the ball would appear next, we could reduce the window and set it to the expected position. In order to calculate the trajectory, the ball's trajectory was determined numerically, integrating numerically the partial differential equations of the flying ball including spin, variable Reynol's number and the sliding and static friction when bouncing on the table. Parabolic approaches were not sufficiently accurate. Whenever the image processing provided a new ball position, the trajectory parameters were adapted so that the trajectory passed through as many measured points as possible with the least possible deviation. Then the numerical integration was recalculated to determine the exact point of arrival and based on this data the return match was determined. This robot was the first to be developed in the Mechatronics Group at the Institute of Mechanics at ETH Zurich. From this group the Institute of Robotics was founded, which was later renamed Institute of Robotics and Intelligent Systems (IRIS). The robot was thus to a certain extent an important building block in the development of robotics at ETH.
@@einarnielsenntbch yet it is capable of tracking / predicting the location the ball will land in real-time! and moving precisely at that location! I am surprised that this was possible in 1988!
1988!!! seriously!!! does that mean we havent come that far in technology as of 2020!
In fact, many things were developed very early, but the approaches were of course different, because the computing power available at that time was not the same.
The system had two VME processor boards, each with a Motorola 68020 processor with 25 HMz. One was responsible for calculating the control algorithms and the other for calculating the game strategy. The image processing was solved purely electronically. The interlaced signal contained two fields of a video frame captured consecutively with a full frame rate of 25 Hz. Each frame were evaluated, resulting in a frame rate of 50 Hz. A binary image was generated from the camera signals and the centre of gravity of the white pixels was determined in the image. In order to suppress interferences, a window could be defined in which the center of gravity determination was applied. So as soon as we knew where the ball would appear next, we could reduce the window and set it to the expected position.
In order to calculate the trajectory, the ball's trajectory was determined numerically, integrating numerically the partial differential equations of the flying ball including spin, variable Reynol's number and the sliding and static friction when bouncing on the table. Parabolic approaches were not sufficiently accurate. Whenever the image processing provided a new ball position, the trajectory parameters were adapted so that the trajectory passed through as many measured points as possible with the least possible deviation. Then the numerical integration was recalculated to determine the exact point of arrival and based on this data the return match was determined.
This robot was the first to be developed in the Mechatronics Group at the Institute of Mechanics at ETH Zurich. From this group the Institute of Robotics was founded, which was later renamed Institute of Robotics and Intelligent Systems (IRIS). The robot was thus to a certain extent an important building block in the development of robotics at ETH.
@@einarnielsenntbch yet it is capable of tracking / predicting the location the ball will land in real-time! and moving precisely at that location! I am surprised that this was possible in 1988!