When will China's AI factory be realized?

 Outside of the technology sector, AI technology is still mostly experimental, with a few exceptions—particularly in the automotive sector—where few factories have begun to adopt it, and where AI technology has been implemented, it is used on a small scale, mainly in areas such as inventory management and inspection. Let's follow the industrial control editor to learn more about the relevant content.

Dayton Horvath, a researcher at Lux Research, a market research firm, said: "AI can support operations such as finite element analysis (FEA) used to build simulation models, and can also handle more difficult problems, such as those with greater degrees of freedom or incomplete data sets;" For example, topology optimization can use AI to create lighter parts with the same or higher strength, and can build more efficient heat exchanger models.

One system that is often mentioned in factory AI applications is robots. Neocortex, the AI ​​robot technology of the American company Universal Logic, originated from the Robonaut research and development project of the space station robot sponsored by the National Aeronautics and Space Administration (NASA). This technology enables the automated system to handle deformable objects, high item variability and parts replacement without the need for fixed facilities.

Hob Wubbena, vice president of Universal Logic/Universal Robotics, said that Neocortex is the AI ​​machine learning module of the company's Spatial Vision 3D software platform, which can work with various actuators, not just robots; the platform can sense the robot's surrounding environment used for tasks such as handling and grasping, allowing the robot to interact and respond to the environment in real time and at high speed. The robot's capabilities include the ability to properly identify and respond to objects of mixed shapes and textures, such as bottles, bags and boxes, with a reliability of up to 99%.

For collaborative robots, human-in-the-loop reinforcement training is key to making robots smarter with the help of machine learning. Erik Nieves, founder and CEO of PlusOne Robotics, said: "Reinforcement learning will have an impact on factory production lines and distribution centers. In the future, every large factory will be a distribution center, even if the people who run the factory have not thought about this. As factories adopt AI technology, these will gradually develop."

Examples of combining AI with industrial robotics equipment

Recently, the industry has also had two collaborative projects to develop AI technology for industrial (including manufacturing) process robots. One of them is "cognitive industrial machines" that can help human operators improve quality control, increase speed and yield, and reduce down time. This is a combination of ABB Ability's cloud-to-edge device cross-industry digital solutions and IBM's Watson Internet of Things platform.

The commercialized system is called Cognitive Vision Inspection System. It combines the AI ​​of Watson supercomputer and the real-time images of production lines captured by ABB system to find defects and send relevant data to the cloud for analysis by Watson IoT platform dedicated to manufacturing industry. Bret Greenstein, vice president of IBM Watson IoT, said that Watson is executed in the cloud, and a subset of it can be executed on the server: "We can run it on edge devices, gateways, usually x86 systems with Linux or embedded operating systems. We are working with Cisco and other manufacturers in this regard."

In addition to supporting machine vision inspection, IBM uses Watson's perception capabilities to interact with operators in a hands-free environment or provide augmented reality tools to assist in diagnosing and repairing equipment. Greenstein said: "We are seeing the adoption of this technology around the world, including in the United States and other markets; AI brings more competitive advantages, including improved quality, safety and productivity, and the manufacture of more sophisticated and complex products."

At the same time, Nvidia and Japanese company Fanuc are also working together to add AI capabilities to Fanuc's industrial control system Field (Fanuc Intelligent Edge Link and Drive), allowing robots in automated factories to operate faster and more efficiently. This technology will apply a series of Nvidia graphics processors (GPUs) and deep learning software to enable AI to be executed in the cloud, data centers, and even embedded in edge devices.

The Field system is linked to CNC equipment, robots, peripheral devices and sensors to optimize manufacturing production through analysis; Murali Gopalakrishna, director of product management for intelligent machines at Nvidia, said Fanuc recently demonstrated three basic applications of AI robots, including grabbing and placing objects, predictive maintenance at the edge, and automated optical inspection with a seven-fold increase in detection rates.

Nvidia's Volta is claimed to be the first GPU architecture built specifically for AI applications, that is, it can support machine learning training; the Volta architecture Tesla V100 GPU is equipped with 640 tensor processor cores and can provide 120Tflops of performance, equivalent to 100 deep learning CPUs

General Electric, a major US manufacturer, is also developing technologies internally that are suitable for its own manufacturing needs and those of other vertically integrated US manufacturers. John Lizzi, head of robotics at GE Global Research, revealed that in addition to software and hardware platforms, GE has also invested in Clearpath Robotics, which excels in autonomous mobile robots, and OC Robotics, which is famous for its "snake-arm" robots.

For some use cases, GE builds robots from scratch, such as devices that can penetrate deep into jet engines for inspection; the company also builds and purchases sensors. AI has become very important in the field of robotics through machine learning, and this technology has become the key to improving robots in three aspects in the future, namely perception, advanced reasoning and dexterity. Lizzi also pointed out that collaborative robots are also a major trend. GE's vision is to develop towards mobile, self-sufficient systems, where humans only need to intervene when dealing with exceptions, and to deploy intelligent robots that can work with human teams.

The above is the introduction about "When will the AI ​​factory made in China be realized in industrial control?" If you want to know more relevant information, please pay more attention to eeworld. eeworld Electronic Engineering will provide you with more complete, detailed and updated information.