High-speed cameras, as represented by various sensors capable of capturing over 1000 data points per second, are widely available. With a large amount of data, one might think it would be useful for robot control. However, for slow-moving robots, such as the most common industrial robots, it is difficult to respond to rapidly updated data every time. In our laboratory, we are researching technologies that can effectively utilize abundant data even when there is a significant gap in “speed.”
Grasping Moving Objects with a Manipulator Using a High-Speed Camera
High-speed cameras possess excellent motion vision, allowing them to capture high-speed phenomena that the human eye cannot detect. In contrast, the movements of typical industrial robots are slow, making it difficult for them to keep up with fast phenomena. In this research, we aim to enable robots to handle high-speed phenomena that exceed their own movement speeds by maximizing the use of high-speed camera measurement data and accurately predicting the future movements of objects.