Can “machine eyes” see society and the future?

Can “machine eyes” see society and the future?

-The people who created machines beyond humans–

“Machine eyes” beyond humans

The sound of the wind is getting louder and louder.
When I think it has grown, the next moment is a light-driving passenger car.

The sound became low, and when I thought I was far away, I continued to drive at high speed under the strong vibration of the truck and sports car.

The technology to capture and recognize license plates by tracking successive cars with different models and different speeds.

Even athletes with excellent dynamic vision can never catch up.

Masaki Suwa, a senior engineer in image sensing technology, created a technology that recognizes objects at a faster rate than humans and simultaneously captures the three dimensions of human space.

Professional Masaki Suwa from the Technology and Intellectual Property Headquarters
It can be said that image detection is “the eye of the machine”.

No matter how powerful the machine is, if it does not have the “five senses” of the sensory organs, it will become a machine that can only move blindly.

Image detection technology is built into equipment and equipment, and plays the role of capturing the “eyes” of objects.

It is said that humans can obtain more than 80% of the information around them visually.
The machine cannot cooperate with the surrounding environment to control itself, nor can it see the constantly changing situation.

For more than 40 years, Omron has developed advanced image detection technology to replace the “human eye” and has supported the development of society in various fields of industry, society and life.

Suwa first participated in the development of 2D character recognition sensors at OMRON.

Recognize the characters and codes in the images taken by the camera, and read the characters and codes (such as the expiration date and food production batch number printed on the product by the factory) to identify the license plate and read the business. card.

This technology has been widely used for business card recognition in smartphone applications.

At that time, the development of road traffic sensors equipped with the industry’s first technology determined the life of the visiting Su city developer.

On the road, by accurately determining the volume of traffic, the speed of the car you are driving, the type of vehicle, etc. And by controlling the signal or displaying the guidance on the traffic information board, the flow of traffic is controlled to avoid jams. . ..

However, 2D image detection sometimes does not adapt to changing road conditions such as shadows from buildings and parallel cars, cars lined up in traffic jams, and headlights reflecting off the road.

“Can we use two cameras to measure distance capturing the same ‘depth’ of the human eye?”
When this concept was born at the end of the development project team discussion, Suwa felt great future and potential.

He continued to search for a method to install two cameras on the same principle as the human eye.
What is needed is durability to maintain normal driving even when exposed to rain and wind, and high precision that can capture vehicles up to 60m away and achieve precise eye functions.

What bothers Suwa most is the need to achieve such high performance in a limited size.

Humans can immediately process a large amount of information from both eyes, to understand the surrounding situation and easily recognize various three-dimensional objects. Supercomputers are not enough to process the same amount of information.
“It’s hard to feel when the size is small.”

To solve this contradiction problem, Suwa completely reduced the range of information needed from the images captured by the two cameras and improved the accuracy of depth detection.
It is a challenge because it is difficult. Such a developer soul promotes difficult development.

With the development of this technology, cars with a recognition rate of 97% or more can be detected regardless of day, night or weather.

This road traffic sensor is equipped with a mechanism that can receive light through two cameras and calculate the distance based on the difference in the path of entry of the light. This is unique to Omron and cannot be replicated anywhere else.

Identify the base and perform high-precision refinement

Three-dimensional recognition of electronic components and display of color difference
Suwa can be said to be the top engineer of OMRON imaging. He spent most of his time developing it.

This technology is used for three-dimensional reconnaissance inspection to check the shape and joint condition of the “solder” of electronic components mounted on it. . matrix. It is said to be 3D image detection technology.

The board inspection device is a device for high-speed inspection of the soldering state of the board, and hundreds of electronic components smaller than 1 mm square are mounted on the board.

Different from using two cameras such as traffic sensors for 3D image detection, one camera is a camera and the other is a projector that projects a specific pattern onto the whiteboard.

I think that three-dimensional objects can be recognized by the subtle deformation of the projection pattern, but it is not easy to re-develop them.
Such equipment requires precision and speed.

Developed a 3D image detection function, which can change patterns at an ultra-high speed of 100 times per second, which is not recognized by the human eye.

By numerically controlling the shape and joining state of the “solder” that is critical to the quality of electronic devices, high-quality, high-speed products can already be produced.

Illuminate the electronic component (left) with a specific pattern (right) to recognize the three-dimensional shape
It has been 18 years since Suwa started developing 3D imaging detection.

The technology has finally flourished and the scope of the device has expanded. 3D image detection has become one of Omron’s core technologies.

Omron (OMRON) has created unsurpassed image detection technology and, combined with the advantages of OMRON (OMRON), Omron continues to improve its technology to achieve high performance from small parts.
“It is not necessary to process every pixel of the image captured by the camera.

This is the essence of Omron image detection, which can refine the really important information at high speed and high precision. It is also a place to test the source of feeding, “Wa said.

Today is the cloud era.
Supercomputers can be used to collect big data and calculate and analyze large amounts of information.

However, if the accuracy of the data composing the big data itself is poor, accurate answers cannot be obtained, and if a supercomputer is used to calculate everything, the best time for judgment at this time can be determined. You might miss it.

The core of the Internet of Things is how to extract the reality that is happening in the data.

Omron considers how to operate “wisely” with limited resources (such as the size of the CPU and memory that can be integrated into the device) and the compact, high-performance human eye and brain without large-capacity hardware. We have developed devices with built-in functions.

As a palm-sized device, it is active in cities, buildings, roads, factories, and anywhere in society, and can provide information about the machine “machine eye”.

Palm-sized Image Sensor “Components of Human Vision”

Can “machine eyes” see through thoughts and the future?

By looking into the “eyes of the machine” many problems can still be solved.
How to make the invisible part visible.

Wa’s visit to the city not only tried to perceive “places that are not physically visible”, but also tried to perceive “within humans” and “the future as a timeline”.

For example, it is still difficult to detect physically invisible objects like cars approaching from the other side of the corner or in heavy fog with sensing technology.

On the other hand, the feeling of “seeing” the human interior is also evolving. Omron has developed a facial recognition technology that can automatically recognize human faces.

As a technology that supports autonomous driving, people have high expectations about it, but in the future, you will not only be able to recognize faces and movements, but also express facial “fatigue”.

I want to develop techniques for “looking in”, such as “I want to know if I am sleeping” or “I want to know if I am sleeping.”

On-board sensors can estimate the driver’s attention based on the driver’s behavior and facial expressions
The “perception of the future” that predicts the future is no longer a dream story.

Make predictions within a few seconds in the near future, such as predicting the entry of vehicles behind and accelerating in the rearview mirror, or predicting and avoiding the movement of approaching vehicles due to narrow roads. Already standing.

Fully forcefully accepted the interview.
“Development is a series of difficulties. But when we overcome obstacles and push products to the world, many people will be surprised and happy. Really imagine that our technology will change society.

This is the driving force for daily development.”

Omron
Develop technologies that predict the future as much as possible, one step ahead of the future, so that the future can bring about a better society.