Headquarters Australia
Beckhoff Automation Pty. Ltd.

Building 4, 163–179 Forster Road
Mount Waverley, VIC 3149, Australia

+61 3 9912 5430
info@beckhoff.com.au
www.beckhoff.com/en-au/

Nov 16, 2026

Vision becomes a machine function

Machine vision seamlessly integrated into control technology

Published in: inVISION 06/2026, TeDo Verlag, https://invision-news.com

Traditionally, machine vision systems have been developed separately from conventional machine control systems. In practice, this separation leads to latency and, in particular, to additional costs during machine set-up. Beckhoff is breaking down this historical barrier with a comprehensive range of EtherCAT Terminals, illumination solutions, and cameras. This is all being made possible thanks to an integrated automation environment in which both vision-specific and traditional machine processes can be designed.

EtherCAT Terminals for virtually any vision task illustrate Beckhoff’s holistic approach, in which all necessary machine functions – including vision – are implemented within a single integrated system. These I/Os significantly improve the integration of machine vision into the machine control system. Among them is the EL2574 digital output terminal, a pixel LED terminal which drives a string of LED pixels with integrated chips connected as strips or matrices. Typical applications include chaser lights and animations, position indicators, or the display of machine status. A string of LEDs can use green light to indicate that a product has passed a test during the manufacturing process, for example. EL2564 4-channel digital output terminals control LEDs with integrated series resistor via pulse width modulation in voltage-driven mode. Two versions are available: the EL2564 for common anode RGBW LEDs (up to 4 A per channel) and the EL2564-0010 for common cathode RGBW LEDs (up to 3 A per channel).

Beckhoff's vision products also support end-to-end EtherCAT communication for seamless integration into the control system.
Beckhoff's vision products also support end-to-end EtherCAT communication for seamless integration into the control system.

The EL2596 1-channel EtherCAT Terminal is designed for high-precision time-based voltage- and current-controlled control of LEDs (up to 24 V). The LEDs can be operated at up to 1.2 A in continuous mode or up to 3 A in pulse mode. In pulse mode, a minimum duration of 25 µs can be achieved. The EL2596-0010 1-channel LED output is also available, which can drive LEDs with up to 48 V. Moreover, the EL2596-0010 supports a maximum current of 1.2 A in continuous light mode and 3 A in pulse mode. All terminals in the EL2596 series support synchronized operation of multiple LED outputs using XFC technology. EtherCAT distributed clocks provide the common time base used by XFC functions such as timestamping. Distributed clocks are a hardware-based time synchronization mechanism in EtherCAT. They synchronize the local clocks of all network nodes to a common system time with a precision that is usually less than 1 µs. In addition, the EL2596 provides an electrically isolated external output for the trigger signal of a camera.

Beckhoff also offers the right tool for motion applications in the I/O sector: The EL2521 is a simulation encoder terminal that generates an encoder signal synchronized with the controller for driving line scan cameras.

Illumination with high-precision synchronization

Beckhoff's industrial-grade high-performance illumination solutions represent the logical evolution of the EtherCAT Terminals. Three different geometries are available: bar lights from the VIB2000 series, ring illuminations from the VIR2000 series, and planar panel illuminations from the VIP2000 series. These geometries provide support for a majority of vision applications that are relevant not only to machine building. All devices feature multi-color illumination with four color channels: one green, one orange, one blue, and one infrared (850 nm) – each of which can be controlled independently. This channel configuration ensures that the features under examination are displayed with maximum contrast.

The illuminations are optimized for pulse mode. The start time of the pulse can be specified as a timestamp in the same way as on the EL2596 EtherCAT Terminals. This allows the transmitted pulse to be optimally synchronized with the PLC. The strengths of this approach are particularly evident when synchronization of multiple illuminations is essential for the application. The illuminations can be precisely controlled at different points in time in shape-from-shading applications, for example.

Trade show prototype for highly precise synchronization of camera, illumination, and rotating motion axis
Trade show prototype for highly precise synchronization of camera, illumination, and rotating motion axis

Industrial-grade cameras for direct machine integration

Beckhoff offers GigE Vision®-compatible cameras with an integrated EtherCAT interface. The VCS2000 series can be configured equally well using EtherCAT or the GigE Vision® protocol. However, it is worth noting that parameterization via EtherCAT makes it easy for manufacturers of production machines to configure the camera using startup commands. In addition, replacing the camera is much easier, since the new camera is configured entirely via EtherCAT.

Image data is always transmitted via the built-in 2.5-Gbit Ethernet interface, regardless of how the camera is connected.

Software, real time, and engineering

Beckhoff's TwinCAT Vision software integrates image acquisition and image processing functions into the TwinCAT Engineering environment that machine builders and control programmers are familiar with. TwinCAT Vision is certified to the GigE Vision® standard version 2.0. This means that GigE Vision®-compatible cameras can be used for image acquisition with TwinCAT Vision. Image acquisition and the associated camera parameters are easily configured during the design phase, and the parameters are written to the camera when the PLC starts up.

TwinCAT Vision offers a comprehensive library of common image processing functions. These include, for example, object detection, measurement (including monitoring of processing time), and the presentation of results. The image processing chain is structured in the engineering environment in a way that allows for version control and is understandable to PLC programmers. Images that are intermediate results of the image processing chain can be displayed in the engineering environment while the PLC is running. In addition, recorded images – just like those captured by a camera – can be used in the PLC. This feature makes setting up the image processing chain much easier, as the results of the image processing functions can be verified even if a camera is not connected.

All image processing functions in TwinCAT Vision are executed within the TwinCAT real-time context. Their execution is therefore strictly deterministic. Watchdog functions can be used to limit image processing functions whose runtime depends on the image content to a maximum execution time. Some of these image processing functions return partial results even if execution is aborted.

For PLC programmers, TwinCAT Vision is an extension of their usual, familiar tasks in the context of control system development. Traditional image processors are usually engineered in a very different way. TwinCAT Analytics Vision provides a convenient way of developing and testing vision algorithms on a simple graphical interface. It is a code generator that generates PLC code at the touch of a button to implement the configured image processing chain. This code can be used directly in the PLC and also allows for version control. Changes can therefore be tracked easily and transparently, making the code significantly easier to maintain.

AI made easy for vision too

Multiple Beckhoff bar lights can be used to create a flexible ring illumination
When precisely synchronized via EtherCAT, multiple Beckhoff bar lights can be used to create a flexible ring illumination that can be lit in a ring configuration or with individual segments controlled at different times, entirely in line with prevailing requirements.

The ease with which vision data can be handled in TwinCAT is also an important factor in ensuring that artificial intelligence can be used effectively in industrial applications. Although training AI models is typically a time-consuming and complicated process, Beckhoff’s TwinCAT Machine Learning Creator makes it easy to create AI models without requiring the specialized knowledge of a data scientist. The trained models can then be easily exported in standardized ONNX format.

Beckhoff offers various modules for inference: with TwinCAT Vision Neural Network, inference takes place in real time with CPU support. The TwinCAT Machine Learning Server is typically used for more complex tasks. Although this service does not run in a real-time context, it does offer GPU support for inference.

Vision meets motion

The seamless integration of Beckhoff Vision described above really plays to its strengths in machines with multiple axes of motion. The TwinCAT XFC Library can be used to calculate a timestamp corresponding to a specific position for NC axes or, in the case of rotary axes, a specific angle. This allows the camera, illumination, and EtherCAT terminals to be precisely synchronized with the motion. For unambiguous assignment of image data, the timestamp of the image capture can be embedded in the metadata of the imported image and read out with TwinCAT Vision.

A trade show prototype that was showcased at Vision 2026, among other events, demonstrates this highly precise synchronization of camera, illumination, and NC axes. The prototype features a Beckhoff VCS2000-0200 camera with a small region of interest that is typically only a few lines (8 to 64) in height. The lines are scanned in rapid succession at a rate of more than 1 kHz, combined into a single image in the camera, and then sent to the control system as a complete image. The number of individual lines that are combined to form an image can be set on the camera. The controller then uses TwinCAT Vision to reassemble these partial images.