24-month Festo Didactic warranty → www.festo-didactic.com 136 137 Machine learning in image processing The “MPS IoT Retrofit” package deals with machine learning based on neural nets (also referred to as “deep learning”), one of the most prominent sub-disciplines of artificial intelligence. As the underlying application scenario “computer vision” has been selected, therefore the hardware contains a singleboard computer equipped with a HD camera to record images to be analyzed by a neural net afterwards. With respect to the software, a variety of python programs are available. The frontends of almost all tools are web-based, allowing for an access via mobile devices like smart phones, tablets, laptops, etc. Due to the Wi-Fi hotspot the device offers a wireless remote access. The system comes ready to run so students can start first experiments right away. The key feature of the “MPS IoT Retrofit” package is the fact, that students learn the most important topics in the domain of image processing with machine learning in an easy way. Not only the main learning schemes like supervised/ unsupervised learning are discussed, but also the most prominent applications in the computer vision domain – i.e., image classification, object localization, and multi-object detection – are introduced and discussed by means of a series of practical experiments. Students can distinguish apples from lemons or tools from shoes etc. One potential task when integrated in a learning factory is to check the filling level of slides containing workpieces by applying machine learning techniques. Otherwise, all kind of objects can be detected and localized. Furthermore, powerful neural net architectures like so-called convolutional neural nets are being used. Artificial intelligence in application MPS IoT kit Machine Learning Focus and trending topics I4.0 > Artificial intelligence in application The entire software is well- documented and allows students to perform their own computer vision experiments outside the learning factory. Prior programming knowledge is not required. The Courseware encourages students to transfer their knowledge to new applications. Learning contents – Artificial Intelligence/Machine learning based image processing – Practical application of (convolutional) neural nets/deep learning – Supervised and unsupervised learning – Computer vision (image classification, object localization, multi-object-detection) – IoT Retrofitting of legacy systems Benefits – The device can either be used stand-alone or be integrated in an existing learning factory – Students have the option to apply the algorithms to new objects and images – Focus on practical application of AI/ML solving real-world challenges Main components – Single-board computer with HD camera – Ethernet cable – HDMI cable – Power supply Technical data – Power supply: AC 110/230 V, 1 A – Dimensions (W x D x H): approx. 200 x 200 x 600 mm MPS IoT kit Machine Learning 8158958 MPS IoT kit Machine Learning (Classroom set = 8 devices) 8158957 IIoT and Retrofitting IoT Gateway Focus and trending topics I4.0 > Artificial intelligence in application IoT Gateway 8172682 The connection of components at the production level with servers at the IT level of a company or with the cloud offers a wide range of new usage possibilities, e.g. monitoring production with a smartphone via the internet. Gateways play an important role in interoperable communication via standardized data exchange protocols and also frequently offer functions for decentralized data management. The IoT Gateway connects production level devices to the Industrial Internet of Things (IIoT). It has a network connection for the device side, one for the cloud side and a hardware switch to control read and write authorization. The gateway offers a web interface with configuration options, including – Network configuration including DHCP client – NTP client for time synchronization – Device management – MQTT broker settings The gateway is able automatically to find known device types such as the Festo Didactic energy measurement box in the network. The information for pairing the devices is stored in a signature file. After the devices have been paired, referred to as onboarding, the data is automatically retrieved cyclically and forwarded to an MQTT broker. Your own signature files can be created and imported, meaning that your own device types can be found, coupled and read out via OPC UA. The graphical development environment installed on the gateway, NodeRED, enables edge computing functionalities, i.e. data processing at the boundary between the local network and the cloud. A wide range of signal sources can be integrated using library elements, e.g. via the protocols OPC UA, Modbus TCP or REST API, signals can be pre-processed using function blocks or JavaScript code, dashboards can be set up for visualization and signals can be output to server services such as MQTT, MySQL or cloud services such as Siemens MindSphere or Microsoft Azure. The gateway can be installed with the supplied accessories e.g. in CP Lab carriage or on the NetLab EduTrainer and connected. Scope of delivery – IoT Gateway – Connection cable 24 VDC to 4 mm safety plug – 2x network cable – Mounting accessories – Training documents with example scenario
RkJQdWJsaXNoZXIy MzE0OTAwNw==