The two sensors for the fully automatic "reach into the box" extend the successful Intellipick 3D products from Isra Vision. Particularly robust and with high speed, the two sensors meet even difficult production requirements such as harsh environments and the shortest cycle times. Equipped with embedded technology, WLAN and the OPC/UA communication protocol, they are also optimally equipped for networked production. With the embedded PC, the design of both sensors eliminates the need for extensive cabling between the camera and PC, allowing data transmission to achieve maximum speed.

The newly equipped Intellipick-3D-Pro is suitable for difficult lighting conditions with its multi-line laser illumination and optional software extensions. It has been specially designed for the most demanding challenges on production lines. The laser illumination enables detailed image capture with a high depth of field, even at large distances from the container, in incident ambient light and under harsh conditions. Parts as small as 15 millimetres are reliably detected and gripped, even in complex positions. The system detects dirty surfaces as well as shiny or matt components. Collision avoidance automatically plans the optimum gripping path and enables trouble-free operation. Thanks to a software extension, the system tolerates different container positions and types.

With quad-camera technology, the Powerpick-3D achieves new top speeds for scanning times in the millisecond range. For this multi-stereo process, four integrated cameras capture the components to be gripped in the container. Powerpick-3D automatically creates an optimized gripping sequence and calculates the corresponding robot paths. As various camera angles are available thanks to the quad-camera equipment, the sensor's multi-view images remain reliable even if there are shadows in the sensor's field of vision or light reflections on the component surfaces.

All bin-picking solutions are based on point cloud technology, which enables them to fully capture the objects inside a container. The various sensors extract the components to be picked from this - initially unstructured - information using a CAD template. As the system learns new part geometries quickly and efficiently using a CAD data set, the number of components that can be detected is virtually unlimited. pb