The perception camera “mvPerCam” is an optical system that can perceive objects in its environment. The features and the basic principles are based on human vision in connection with the human brain. i.e. the human brain interprets the scenery instantly and recognizes a chair, a table, a PC and not a bunch of gray scale pixels. The objects are already known, because we learned throughout life what to do with them. We recognize high contrast differences, like edges and structures, very well. We also recognize the distance between the objects. But how does this work with a grasshopper in grass? – With movement. We can recognize brown mushrooms in the forest because of the different movements between the mushroom in the foreground and the background, which in turn are due to the different distance between foreground and background. Nature is the greatest teacher. The task of the mvPerCam is to perceive known objects. For this reason you have to describe the general characteristics of the objects that the application should recognize.
The mvPerCam shows all objects with these characteristics in video real-time (30 Hz) at full resolution (1024 x 1024 pixels). It will return the position (3D) and the motion vector (6D) of the recognized objects as well as the size, the orientation in space and the perfect pick point. At the same time the camera can differ between closely positioned packages. The detection area starts from 250 mm and reaches 2500 mm; best for object tracking or as an additional vision part of a robot, which first looks at a whole pallet and then, more closely, at the single packages. Hits or repeated shaking of the transport system will not have any effect on the mvPerCam because it uses a continuous self-calibrating system.
The mvPerCam is suitable for applications in the logistics and automotive areas. For example the camera determines sizes and volumes of moving or static objects and returns the necessary data for pick & place tasks in complex scenes. Because of its robust design and fast detection speed, the camera is also suitable for controlling robots in real-time.
