Door releases artificial intelligence module for Simatic control system • The new TM NPU module
• The new TM NPU module integrates an artificial intelligence (AI) chip for Simatic S7-1500 controllers and ET 200MP I/O systems
• Use neural networks to analyze data such as video, sound or CPU
• Machine learning algorithms can be used in the production process
• Effectively perform sorting/placement applications or quality inspection based on (human) expert knowledge
Siemens introduced a new module with integrated artificial intelligence (AI) chips for the Simatic S7-1500 controller and ET 200MP I/O system. The Simatic S7-1500TM Neuroprocessor (NPU) uses the Intel Movidius Myriad X Vision Processor (VPU) to enable efficient processing of neural networks. The new module is equipped with a USB 3.1 interface and a Gigabit Ethernet port to obtain the functions of the trained nervous system through the SD card. The sensor data and the data from the CPU program are processed on the basis of the neural network. With machine learning algorithms, applications such as visual quality inspection of production plants or robot systems with image guidance will be effectively implemented. These behaviors become more efficient and more like human experts. With the introduction of this module, Siemens will integrate more future technologies into industrial applications.
The VPU installed on the Simatic S7-1500 TM NPU features Intel's new Myriad X VPU chip and a dedicated hardware accelerator with a deep neural network structure. Myriad X, which integrates an image processing unit and a neural network computing unit, is a pioneer in computer vision applications. Embedded Intel chips enable new applications in industrial automation based on trained models to accelerate image processing and fast local data evaluation.
Users can connect compatible sensors, such as cameras or microphones, sensor data and information from CPU programs, on the integrated interface of the S7-1500 TM NPU module, using neural networks for processing, and the results of the processing are evaluated in the CPU program. When using traditional image processing technology to identify workpieces, the data of each workpiece must be set extremely accurately. The application learning algorithm analyzes the identified image data, and the setting of the workpiece data can be more flexible to a certain extent. Open artificial intelligence frameworks such as Tensorflow are used here.
The advantages of the Simatic S7-1500 TM NPU in the sorting/placement application scenario are reflected in the fact that the mobile robot must be able to identify, sort and place the components that are randomly placed in the box. In addition, additional value will be brought to the quality inspection. For example, the neural network will be continuously trained by using images or data collected by the networked camera. The module will have the knowledge of product or process consistency, color and quality parameters like an expert.
Background Information:
Artificial intelligence has many uses, and its application in the industrial field can greatly reduce the workload of programming and engineering development. It can make control logic more flexible and agile to respond to changes in environmental conditions, and more flexible and precise organization of production processes.
Siemens' Future of Automation project demonstrates Siemens' vision for the future of automation and the future of artificial intelligence in the Totally Integrated Automation portfolio. This means a flexible solution from the field layer, controller layer and edge layer to the cloud. It also means that users can scale the scale of artificial intelligence solutions based on the environment and target applications: for example, the field layer needs to respond quickly, while at the cross-device and shop floor levels, massive amounts of data need to be processed and the corresponding computing power is required.
