Select your location:

Location

Select your language:

KUKA Robot Guide
Find case studies and robots for your industry and desired application with the new KUKA Robot Guide!
Start the KUKA Robot Guide now!

iiQKA.AI Vision: Industrial image processing for robots

iiQKA.AI Vision is an integrated machine vision software for manufacturing and industrial robotics that combines 2D and 3D vision, AI-powered image processing, and robot integration in a single solution. With intuitive drag-and-drop engineering in iiQWorks, users can rapidly configure use cases such as bin picking, assembly, quality inspection, and depalletizing. Running directly on the robot controller, the platform reduces system complexity, minimizes hardware requirements, and enables low-latency processing for scalable automation.

2D and 3D vision with AI-based object detection and localization

The software detects, classifies, and locates components in 2D and 3D. It can process image data or point clouds for applications involving complex geometries, disorganized components, and changing environmental conditions. Depending on the application, the result can include the position, orientation, and gripping information required for handling.

Processing takes place on the KUKA AI Extension Platform, an integrated computing unit based on NVIDIA technology. The computing unit executes image-processing and AI algorithms directly within the controller. This avoids the need for an external industrial PC for the integrated processing function and reduces the number of interfaces between the vision system and the robot.

Key benefits: Efficient, integrated and ready for immediate use

  • Accelerate machine vision deployment for robotics

    • Visual configuration via drag-and-drop in iiQWorks
    • Templates for typical applications
    • Guided workflows from setup to start-up
    • Engineering without traditional programming

  • Reduce hardware costs and system complexity

    • No external IPC hardware required
    • Integrated processing on the robot controller
    • Fewer interfaces and system components
    • Lower integration and maintenance effort

  • Reliable 2D and 3D machine vision performance

    • High accuracy in 2D and 3D object detection
    • Automatic localization and grasp point calculation
    • Reliable even with varying components and lighting conditions
    • Ensures collision-free path planning

  • End-to-end vision workflow from engineering to operation

    • Seamless integration into iiQWorks and robot controller
    • Standardized workflow from configuration to deployment
    • Centralized management and monitoring via web-UI

AI Vision for Robotics: 2D & 3D Object Detection, Bin Picking and Grasp Point Calculation

iThe machine vision software of KUKA combines AI vision, machine vision, and robot vision in an integrated solution for industrial image processing. With 2D and 3D vision, the software enables efficient automation of of manufacturing processes such as bin picking, inspection, assembly, and depalletizing.

Machine vision applications in manufacturing

3D bin picking

  • Creates and processes 3D point clouds for object recognition
  • Precisely locates objects and derives gripper points from CAD data or trained logic
  • Automatically calculates collision-free and stable gripping strategies

2D and 3D machine vision

  • Detects workpieces and their position in front of and inside the machine
  • Ensures reliable handling without a precisely defined position
  • Ensures stable loading and unloading processes

3D depalletizing

  • Detects positions, positions and alignment of products
  • Dynamically adapts the gripping strategy to different stack configurations
  • Ensures stable workflows even in the case of varying patterns

2D and 3D conveyor tracking

  • Detects moving objects in real time
  • Synchronizes the robot motion with the conveyor speed
  • Continuously calculates precise positions for gripping while in motion

2D and 3D positioning and robot guidance

  • Detects parts in 2D or 3D and determines exact coordinates
  • Guides the robot to the end point with exact coordinates in real time
  • Automatically compensates for position deviations
  • Reproducible processes

Inspection and quality control

  • Provides reliable product inspection with automated detection of surface features, colors, labels, barcodes, text
  • Performs presence/absence verification and quality control tasks
  • Delivers consistent inspection results through precise image processing

Machine vision software

Machine vision software captures and analyzes visual data. Depending on the application, it can combine traditional image-processing methods, AI-powered algorithms, and robot integration. Rule-based functions can be used for clearly defined inspection tasks, while AI-based functions support object recognition and other applications involving greater variation.

iiQKA.AI Vision combines machine vision and robot vision in an integrated solution. Applications are configured visually and deployed to the robot controller, supporting end-to-end industrial vision systems.

Vision sensors, cameras, lenses, and lighting

The vision sensor captures the images or 3D data required for the application. iiQKA.AI Vision supports a range of 2D and 3D vision sensors. Camera type, lens, lighting, and mounting position should be selected according to the workpiece, required field of view, working distance, surface properties, and process.

There is no single camera or lighting setup that is best for all machine vision systems. Reflective, dark, transparent, textured, or changing surfaces can require different imaging conditions. A suitable setup must provide image data that the software can analyze reliably.

Computing unit and AI acceleration

The computing unit is an integral part of iiQKA.AI Vision. It executes image-processing and AI algorithms locally within the controller and supports the integration of vision results into the robot process.

An AI board extension accelerates computationally intensive AI functions. GPU-based technologies, including NVIDIA technology, support demanding 3D data processing and AI-powered image analysis directly in the controller.

Engineering in iiQWorks (configuration, simulation, start-up)

Engineering is performed in iiQWorks within a seamless, visually guided environment – from configuration and simulation through to start-up. Vision applications are created and directly integrated without conventional programming.

  • Visual configuration: Drag-and-drop instead of traditional programming
  • Template-based entry: Quick setup of typical Vision applications
  • Data & calibration: Camera setup, recording, and system calibration
  • AI-supported models: Training and optimization directly within the engineering workflow
  • Seamless integration: Transfer of results to the robot process
  • Simulation & start-up: Validation and fine-tuning within the same environment

System integration with the robot and controller

iiQKA.AI Vision combines sensors, software, computing, and the KUKA robot controller into an integrated system. Vision results are incorporated into the robot application and can be used for motion and handling decisions.

This integration reduces the need to connect separate systems and provides a common workflow from configuration through operation. Application-specific integration requirements still depend on the robot, sensor, workpiece, and process.

Start your vision project

Configure iiQKA.AI Vision for your industrial image-processing application in KUKA Shop, including the selection of suitable cameras and lenses. KUKA can also support projects from the feasibility study through to stable operation, depending on the project requirements.
Request personalized advice Configure your machine vision solution

FAQ about industrial machine vision and robotics

What is a machine vision system?

A machine vision system uses cameras, sensors, lighting, image-processing software, and computing hardware to capture and analyze visual information. In manufacturing and industrial automation, it can detect, locate, inspect, classify, and track parts, providing real-time data to machines and robots for automated decision-making and process execution.

What is AI machine vision, and how is it different from rule-based vision?

Rule-based machine vision uses defined image-processing functions and criteria for clearly specified tasks. AI machine vision uses AI algorithms and trained logic for tasks such as recognizing objects or handling greater variation in appearance and position. The appropriate approach depends on the workpiece, process, data, and application requirements.

iiQKA.AI Vision combines traditional machine vision functions, AI-based functions, and robot integration in one solution.

When 2D vision is suitable?

2D Vision is suitable for applications in which components are positioned in a defined manner and arranged flat. It is primarily used for traditional image processing, inspection, and positioning based on image data, and is particularly efficient when component positions are constant and cycle speeds are high.

Typical applications of 2D Vision:
  • Pick & place on a conveyor belt or tray
  • Surface inspection (e.g., scratches, color, defects)
  • Barcode, OCR, and label verification
  • Presence/absence checks and quality inspection     

 

Advantages of 2D vision in industrial automation:?
  • High speed and short cycle times
  • Cost-effective integration
  • Proven and robust technology for structured processes

This makes 2D vision particularly suitable for standardized applications with a defined component layout where depth information is not required

When 3D vision is suitable?

3D Vision is used when components need to be spatially captured and handled flexibly—especially in cases involving complex geometries, unordered components, or variable positions and orientations in space. Typical applications include, for example, bin picking and precise position determination in 3D space.

This makes 3D vision particularly well-suited for complex and unstructured scenarios in which the exact position and orientation of a component is critical for a safe and reproducible process.

Typical applications of 3D vision:
  • Bin picking (unordered, stacked components)
  • Depalletizing and handling mixed parts
  • Gripping complex geometries
  • Applications with variable position, height, or orientation
  • Precise positioning and robot-assisted assembly

 

 When is 3D vision absolutely necessary?
  • Components are scattered or stacked
  • Height, position, or angle are relevant to the process
  • A safe grasping position in space (including orientation) must be determined

What lighting setups work best for industrial machine vision?

Camera and lens selection depends on the workpiece, field of view, working distance, required image detail, sensor position, and application. The lens and camera must provide suitable image data for the intended inspection, positioning, or handling task. Lighting and mounting should be considered at the same time, because they affect the quality of the captured data.

How long does it take to implement a machine vision application?

Implementation time depends on the application, workpiece variation, sensor setup, integration requirements, and complexity. Visual configuration, templates, and guided workflows can reduce engineering effort, but the application still needs to be configured, calibrated, validated, and started up for its specific operating conditions.

Is iiQKA.AI Vision suitable for users without extensive vision expertise?

The solution is designed to support users without in-depth vision expertise through visual configuration in iiQWorks, guided workflows, and templates for typical applications. The required engineering effort still depends on the application and its technical requirements.

Can iiQKA.AI Vision be integrated into existing automation systems?

Yes. iiQKA.AI Vision machine vision software integrates with industrial automation systems by combining image processing, AI-based analysis, robot control, and application execution in a single environment. It supports seamless integration between the vision system, KUKA robot controller, and robot application. Specific requirements depend on the robot, sensor technology, workpiece, and production process.

Is image data processed locally or externally?

Image processing and AI processing are performed locally in the integrated computing unit within the controller. This supports low-latency integration and avoids relying on an external processing system for the integrated vision function.

When are standard software, turnkey systems, or custom solutions appropriate?

Standard software is suitable for common vision applications and can support efficient deployment with limited engineering effort. Turnkey systems combine predefined hardware, software, and integration services. Custom solutions are appropriate when requirements are highly specialized and require a different level of adaptation.

With iiQKA.AI Vision, many industrial vision applications can be implemented as an integrated standard solution while being configured for the requirements of the specific application.

Software solutions for efficient automation