The industrial machine vision market is growing at a record pace. According to the latest statistics from the Association for Advancing Automation (A3), the first six months of 2021 saw an 18 percent increase to $1.5 billion over the same period in 2020, which is the best start to a year on record. And those numbers aren’t expected to decline. Why the increase?

Machine vision systems significantly improve industrial processes and help drive productivity, efficiency, and quality while reducing costs in markets such as automobile manufacturing, food and beverage production/processing, and logistics. Imaging technologies have taken center stage as the main drivers of many advanced applications in automation. It is no longer a question of whether a system should utilize vision and imaging. Rather, these technologies are frequently required to achieve success. Machine vision also plays a crucial role in the broader automation landscape by advancing productivity relative to concepts in smart factories and Industry 4.0, such as AR/VR, industrial internet of things (IIoT), robotic guidance, and big data analytics.

According to Dr. Chris Yates from Vision Ventures, the key drivers advancing the continued implementation of machine vision technologies are an increased awareness of their capabilities and value; decreasing component, software, and engineering costs; broader technology compatibility and interoperability; and greater focus on  ease-of-use.

Trends In Machine Vision Components

Here are a few categories to keep an eye on that are resulting in innovations—and ongoing interest—within machine vision components:

  • 3D Imaging Continues to Advance: 3D imaging has become a mature and robust part of the machine vision market, featuring both new and updated components and systems for critical advanced automation tasks, including metrology, inspection and guidance. Use cases are expanding with the advancement of the reliability, precision and ease-of-use of this type of imaging. A key type of imaging system in this category is the scanning laser profilometer (3D profiler). This device uses laser line triangulation to acquire and create a high-precision profile of the surface of a part, usually with the sensor or part in motion. While many companies offer competing products, one new implementation of this type of imaging comes from Automation Technology GmbH. Its MCS series of modular sensors allows user configuration of the physical layout of the camera and laser line generators. This arrangement adds flexibility in implementation.

    An advance in existing technology from Cognex, the In-Sight 3D-L4000 features new speckle-free blue laser scanning and a broad range of 3D analysis and measurement tools, all implemented within the familiar In-Sight spreadsheet environment. A completely different emerging entry in this category is the Saccade Vision MD 3D imaging system. This device does not require part or sensor motion and can automatically scan a field of view from multiple directions and with multiple variable resolutions within a single image. The new Flash sensor from Teledyne e2v features higher-speed, specialized capabilities specifically targeted for advanced laser scanning systems.

  • Camera and Interface Improvements: Demands for higher-resolution imaging and increased process throughput drive the need for advanced and high-speed machine vision camera components. Supporting high frame rates with large-data images further requires high-speed interfacing between the camera and processor. High resolution and frame rates are increasingly available in imaging sensors for machine vision, driving new camera offerings. Emergent Vision Technologies utilizes a GPixel 103MPixel CMOS sensor in its new Zenith grayscale/color camera. To optimize the sensor’s available frame rate, the Zenith uses a 100GigE interface. This emerging technology in machine vision provides 100 times the speed of basic GigE connections.

    Other interfaces used in machine vision, such as CoaXPress (CXP) and Camera Link HS (CLHS), have evolving standards for transfer rates that also target higher-speed cameras. CXP frame grabbers that support CXP-over-Fiber include the Euresys QSFP+ board and CLHS, which is already capable of 100G over a 4x25G connection, with work currently in process to provide a solution standard capable of 50G.

  • Lens Developments: Lens technologies continue to advance to keep up with the demanding imaging requirements of evolving automation applications. Important features include expanded image format capabilities that support the larger physical sizes of new high-resolution sensors, such as the 1.4" format megapixel MPT series from Computar; optics that provide high-quality images when using both visible and nonvisible illumination wavelengths such as shortwave infrared, as in the Computar ViSWIR series and Kowa’s VIS-SW lenses; and embedded motorized or liquid lens focus control, available, for example, in Edmund Optics’ TECHSPEC LT series and Computar’s LensConnect lenses.

  • Embedded Systems for Deep Learning: As machine vision systems leveraging deep learning (DL) techniques continue to show promise in several different types of inspection applications, a wide range of components and software for implementing DL inspection has emerged. Some of the most recent are cameras and computing systems with onboard (or embedded) processing for deep learning tasks.

    The NEON-2000-JNX smart camera from ADLINK Technology has a GPU-based system with additional FPGA support combined with software to perform edge AI. The Deepview camera from Deepview AI is a self-contained, server-level computing system with imaging that can execute both training and inference for deep learning within a smart camera format. Pleora Technologies offers a computing platform approach intended to facilitate development of AI in machine vision applications.

These are but a few examples of the emerging technologies that are shaping machine vision in automation in manufacturing and other applications outside the industry. For example, it’s helping to automate assisted- and self-driving vehicles, and is also found in agriculture, retail, consumer products, security, mobile robots, and many other areas.

Machine vision also plays a crucial role in the broader automation landscape by advancing productivity relative to concepts in smart factories and Industry 4.0.

New Applications Leverage Advanced Components And Software

New applications tend to arise from the enhanced capabilities offered by the evolving components and software discussed previously. “New” implementations for vision and imaging technology are not sudden changes in the application base but are more the result of a steady progression of development as the technology becomes more reliable and mature.

One of the best new uses for vision and imaging technology is in machine tending. Aided by 3D machine vision systems, robots can grasp randomly oriented parts from bins and present them to automated machining or processing equipment.

This specific application is particularly well suited for the technology, as it typically involves processing only one part type at a time. And the materials and geometries of the parts requiring machining make it easier for robots to locate and grip them. In addition, many 3D imaging software solutions are targeted specifically for bin picking in machine tending, promoting ease of use and flexibility of the application.

While also not brand new, another often talked about technology—hyperspectral imaging—has generated a variety of new applications in wide-ranging markets, including industrial automation, food processing, and pharmaceutical production. For example, hyperspectral cameras can detect chemical content much like a laboratory spectrometer, so they can successfully detect incorrect pills in a drug packaging operation. Similarly, such a system can be used in an automated process to remove spoiled or damaged food products. The technology has thrived outside of industrial automation in the drone-based field analysis of crops to detect drought and disease.

While there are many more examples of these maturing applications, some applications are truly innovative. They are not yet broadly realized but show promise for the future.

Cutting-Edge Applications Emerging

Again, leveraging evolving vision and imaging technologies, applications in direct retail sales are slowly being developed and rolled out. A very visible effort is in Amazon Go stores, where cameras, other sensing devices, and complex AI software track selections and execute cashier-less purchases. This application is being propagated by other businesses and with different paradigms, such as coolers that automatically track outgoing product.

Also cutting edge but potentially ready for growth is the use of vision and imaging technology in automated robotic harvesting of food products in the field. Some small implementations are not yet completely robust, but are showing promise. There even are entire “industrial agriculture” facilities, where robots plant, transplant, and harvest fruits and vegetables in controlled indoor environments.

Regardless of where machine vision and imaging are eing used, it’s more than likely providing great value, improved production, and enhanced quality in many existing and proven applications. As more become aware of its benefits and technologies continue to advance and innovate, expect even greater growth and use cases across myriad industries.