Machine vision has been a rapidly advancing sector, and the integration of artificial intelligence (AI) and machine learning (ML) into the system has propelled its functionalities to new heights.

The evolution of machine vision lenses now incorporates AI and ML to provide superior object recognition and defect detection. “One big movement we see is with liquid lenses being integrated into camera technology,” says Jon Chouinard, vice president, sales and marketing, 1stVision. “Artificial intelligence and machine learning will have benefits in using the controls being built into cameras to adjust the focus using liquid lenses.”

According to Bob Archer, owner, Archer OpTx, the integration of AI into lens design presents an opportunity to determine whether specific attributes should be handled by the optics or processed in the system. “Keep in mind that image processing can’t do much with a bad image, so the optics must at least produce the required image quality,” he says.

Moreover, the importance of quality image acquisition cannot be stressed enough, even in an AI-powered environment. Joe Gugliotti, president, R.J. Wilson Inc., reiterates this. “AI and deep learning technology may help make a poor-quality image more usable, but it’s an inefficient way to utilize software processing power,” he says.

Notable Trends in the Field

Liquid lenses are becoming more integrated with cameras and manufacturers are providing increasingly comprehensive solutions to customers, experts say.

Archer adds that the industry is becoming increasingly environmentally conscious, valuing the thermal sensitivity of lenses and ensuring they withstand harsh environments. He also emphasizes the importance of lighting, terming it as critical as the imaging side.

Gugliotti identifies the trend of cameras moving towards higher resolution sensors with smaller image pixels, stressing the need for lenses to evolve simultaneously and offer support to the minute pixel sizes.

The Evolution of Machine Vision Lenses

Chouinard notes that lens manufacturers have been quick to adapt to changing demands by designing lenses suited to smaller pixel pitches, thereby resulting in improved lens performance.

Lens design has advanced dramatically, Archer says, especially the industry's newfound ability to use unique surfaces and break away from the traditional spherical ones. Despite the evolution, “the lion’s share of the applications out there still use the same old ‘glass elements stacked in a lens barrel’ technology,” he says.

The industrial lens sector hasn’t remained untouched. “The camera sensor trend is going to be toward squeezing more pixels into the standard format sizes available, leading to smaller image pixels,” Gugliotti says. “Lenses must keep up with these advances.”

As the world of machine vision lenses witnesses an integration of AI and ML, the subsequent enhancements in image processing, object recognition, and defect detection are evident.

Machine vision lenses play a pivotal role in diverse industries, from smartphones to drones, wearables, and beyond. As technology scales down and the trend towards miniaturization accelerates, how does it affect the design and application of these critical lenses?

Miniaturization and Machine Vision Lenses

According to Jon Chouinard, vice president, sales and marketing at 1stVision, the path of miniaturization in lenses is generally not a core focus in machine vision. Instead, the emphasis is on creating standard off-the-shelf lenses tailored for sensors used in the industrial imaging market. “Manufacturers are focused on compact design but have to balance cost and performance,” he says.

Bob Archer, the owner of Archer OpTx, has a slightly different perspective. He observes a trend where “lenses are getting smaller and smaller,” and notes the reduction in the number of elements needed for the optical system. The question then becomes whether these smaller lenses are a product of new manufacturing techniques or if the drive towards miniaturization is fueling manufacturing innovations. As Archer points out, the answer is a bit of both, highlighting a symbiotic relationship between manufacturing techniques and lens design requirements.

Joe Gugliotti, president of R.J. Wilson Inc., points out that the size of machine vision lenses is more influenced by their mounting configuration, such as C-mount or F-mount, than the direct benefits of miniaturization. However, he acknowledges the potential of utilizing optical design concepts from miniaturization as industrial sensors become smaller and more pixel-dense.

Types of Lenses in Machine Vision and Their Applications

There are various types of lenses used in machine vision, each with its unique advantages and application areas.

• Fixed Focal Length Lenses: These lenses have a set field of view. Chouinard mentions that they are most commonly used when the application has a fixed perspective and set distances. In contrast, Archer points out that fixed focal length lenses are ideal when the imaged field’s size is known and won’t change.
• Varifocal Lenses: These lenses allow for multiple fields of view, offering versatility. Experts say that varifocal lenses are suitable for applications that lack a predetermined setup, like logistics and sports analytics. However, Gugliotti notes that in his 36 years, he hasn’t used a varifocal lens due to concerns of changing the optical path and the need for recalibration.
• Telecentric Lenses: These are ideal for high-end applications like measurement and gaging. They provide a direct view across the entire field, ensuring accurate gaging even for round parts or parts with a rounded edge, Gugliotti says.
• Fisheye Lenses: These offer a wide field of view but often come with distortions. Some experts suggest that fisheye lenses are apt for surveillance and autonomous vehicles, while others see them as useful in specialized applications such as inspecting internal threads or contamination within a container.

The push towards miniaturization is reshaping numerous industries, influencing design decisions and fostering innovation. While miniaturization may not be a primary driver, the ripple effects of this trend are noticeable, guiding the future of lens designs and their applications.