Picking a part from a disordered bin requires more than recognising its type: the robot needs to know where it is and where to grip it. KUKA’s iiQKA.AI Vision, introduced on 8 October, links those tasks within its controller.[1][2]
The software separates objects from their background. It estimates positions using images or point clouds—sets of points describing the scene in three dimensions—and sends gripping information to the robot. For bin picking, KUKA says gripper points can come from digital part designs or trained logic.[1][2]
Processing runs locally on KUKA’s AI Extension Platform, a computing unit inside the controller based on NVIDIA’s Jetson Orin module. Image processing needs no cloud connection or separate external vision computer.[1][2]
Engineers still choose the camera and configure the application. iiQWorks brings camera calibration, model training and simulation into a graphical workflow. These facilities connect the scene a camera sees with the movement the robot must make.[2]
Sources & further reading
- iiQKA.AI Vision simplifies industrial image processing KUKA · 8 Oct 2026 · Accessed 20 Oct 2026
- iiQKA.AI Vision: Industrial image processing for robots KUKA · Accessed 20 Oct 2026