KUKA has released iiQKA.AI Vision, an AI-based image processing package that detects, classifies and locates parts in 2D and 3D, then passes position and gripping data directly to the robot. The software runs inside the robot controller rather than on a separate vision PC.
The processing happens on the KUKA AI Extension Platform, an integrated computing unit built on the NVIDIA Jetson Orin module. Image processing and AI inference run locally, which removes the need for cloud connectivity or external computing hardware. KUKA says the arrangement gives low-latency processing from data acquisition to execution. Images are processed and stored on board, so the system can operate without a continuous network connection.
Working from either image data or 3D point clouds, the software calculates object position, orientation and gripper point. KUKA says it handles disorganized parts, complex geometries and changing lighting conditions. Listed applications are bin picking, assembly, quality control and depalletising. For automation engineers, the practical change is a vision cell that needs no extra computer or network infrastructure beyond the controller itself.
Setup takes place in iiQWorks, KUKA's engineering suite, through a standalone vision application with a graphical interface. Drag-and-drop tools, pre-built templates for common applications and guided workflows cover the steps from configuration to commissioning. KUKA says this lets teams build vision applications without in-depth specialized knowledge. A web interface handles centralized management and monitoring once the system is running.
Cameras and compatible components are available directly from KUKA.
