Together, AI and 3D imaging work wonders for manufacturers

With manufacturers seeking to make their operations more intelligent and automated, they increasingly turn to AI, which, when performed on 3D images, can benefit manufacturers in numerous ways.
 
Needless to say, factories are increasingly relying on video and analytics to raise production quality and efficiency. Machine vision, for example, can detect defects in a matter of milliseconds. Impending machine failure can also be recognized by way of video.
 
Yet most camera systems used in manufacturing settings produce two-dimensional images, to which application of AI can produce less favorable results. For example, facial recognition, which is required for access control and time attendance for factory workers, can be flawed when working in 2D. “Facial recognition based on 2D camera can easily be spoofed with high-res photos or 3D masks,” said Benson Lee, Director of Strategic Business Development at LIPS, adding that when it comes to dimension measurement, a 2D-based system requires fixed position with pre-calibration, which can be quite time consuming.
 

Benefits of AI on 3D imaging

 
On the other hand, 3D, which adds depth data to 2D images, can produce better and more accurate results. This is where manufacturers can benefit from LIPS, a provider of end-to-end 3D-based solutions from 3D camera module designs to 3D vision middleware and systems. “Our solutions can perform AI inferences of 3D depth images on the edge to perform industrial applications with great speed and accuracy,” Lee said.

 
For example, in terms of the aforementioned facial recognition and dimension measurement use cases, Lee said: “Our facial recognition system based on 3D camera can overcome spoofing. Our 3D-based system requires no pre-calibration and provides greater accuracy.
 
Other applications include automatic guidance vehicles (AGVs), which are frequently seen in factory settings. AI and 3D imaging combined can direct an AGV away from obstacles, guide it towards its intended destinations and are more cost-effective than other solutions such as LiDAR. Pick-and-place operations can also benefit tremendously from AI and 3D imaging, whereby the robotic arm can pick something from one station and place it on another with accuracy and precision.
 
According to Lee, LIPS’s goal is to distribute their products and solutions worldwide. “However, our current geographic focus is in Japan, Korea, Taiwan, China and South East Asia,” he said. “This is mostly because our products and solutions are industrial-focused, and Asia is the manufacturing hub for the world.”
 
Lee also mentions that selling to countries like Japan and Korea has additional benefits. “The most difficult challenge (of promoting our solutions) is really about identifying SIs and the distribution channels with 3D vision and AI expertise, and also communicating to customers about the benefits of a 3D-based system,” he notes. “So we have started to work with several channel partners and SIs in the countries with more advanced manufacturing such as Japan and Korea. They already saw the huge benefits of 3D-based system and we do not need to convince them. We are working with them to develop a partner model that we would like to take to other countries.”
 
Commenting on the Taiwan market, Lee states that there are both advantages and challenges in pushing LIPS solutions. “Taiwan has an excellent education system with lots of capable people. Taiwan is also deeply rooted in manufacturing economy which provides a great exposure for industrial knowledge,” he said. “I think the most difficult part is the value proposition of Taiwan’s manufacturing capability. Only a small fraction would have means to invest and adopt 3D-based system. However, it’s only the matter of time before they reach out for such solutions since their global competitors will start picking up AI.”