 Hello everyone, my name is Thiago Hayes, and today I would like to demonstrate how to leverage the machine learning core functionality when using the SensorTile.box evaluation kit. The demo will rely on the LSM6DSOX 6-axis IMU, and it will also count on our GitHub repository with a series of application examples that are fully available for you to explore. In addition to those, we are going to use the STBLE Sensor app running in expert mode. For this demo, we are going to utilize the SensorTile.box, which is our flexible plug-and-play module that jump-starts IoT applications together with a mobile app. Here on the screen, I also listed both QR codes for you to actually download the STBLE Sensor app, and even though you may not have the SensorTile.box in your hands, you'll be able to follow along some of the steps that are executed directly to the mobile app interface. I'll give you some time to scan the QR code and follow along. To get started, for this demo, we are going to utilize one of the multiple application examples that are available within our GitHub repository. The link to access the repository is listed here on this slide, and will be made available for you after this session. Additionally, during the demo, I will walk you through the repository and show you where to find all the relevant information for you to be able to test each and every one of the examples on your own. So for this demo portion, we'll start by using the GitHub repository that I was just mentioning in the presentation. The GitHub repository will feature not only a list of all the available parts that are compliant with the Machine Learning Core functionality, but also a series of useful documents to help you understand and get started with our development ecosystem. If you go to the main page, you'll find a quick introduction on what is the Machine Learning Core, the repository folder structure featuring all the details, for example, and also a quick description of all the available files. For the demos that we're going to run today, we're going to use the .UCF files. There are basically the Unico configuration files. In this case, they will be utilized together with the SensorTile.box and the STBLE Sensor app. And by the way, you can also find the C header files available as well within the folder structure. The multiple available examples are segmented by application and are targeting the different devices that we have in the Machine Learning Core family. As a starting point then, as I mentioned before, let's go with the LSM6DSOX6AxisIMU. Within the LSM6DSOX page of the repository, you'll find a wide range of examples from 6D position recognition, activity recognition for mobile, GIM activity recognition, head gestures, vehicle stationary detection, vibration monitoring, and yoga pose detection. For this specific demo, we'll go first through the vibration monitoring demo. In this case, for vibration monitoring, you'll find on top of the files that I described, so the .h and .UCF file, you'll find a quick introduction of what the demo actually does. The required device orientation, in this case the orientation is not relevant, and the different decision tree output values, 0 for no vibration, 1 for low vibration, and 2 for high vibration. And then, we also highlight that in this case you actually generate interrupts through the interrupt 1 pin every time the Machine Learning Core Source Register is updated with a new value. As I mentioned, we're going to rely on the .UCF file for the vibration monitoring example. I have preloaded this file that you can download by accessing the repository into my Android tablet, and from there we're going to reach out to the file by using the STBLE Sensor app. When opening the STBLE Sensor app, the first step is to rely on the Create a New App button. When clicking on this button, you'll find a list of application examples that are part of the entry-level mode of the SensorTile.box. If you scroll down out of the way to the end, you'll find a button called Expert View. Let's click on that button. When opening the Expert View mode, you'll be able to find all the applications that you created using the Expert Mode of the SensorTile.box. In this case, for example, let's create a new application. When we create a new application, the first step is to select a new input. And for this case, since we are using the Machine Learning Core, let's select the MLC Virtual Sensor input block. So you just check the box, set as input, and now you can configure your sensor by clicking on the gear on the right-hand side. From there, it's time to select your .UCF file. So let's go ahead and click on Select a UCF File. And in this case, let's run, for example, a vibration monitoring application example using the Machine Learning Core of the LSM6 DSOX. So I'll select this file that is, by the way, the same file as it is available on the GitHub repository. So clicking on this, now we have the LSM6 DSOX vibration monitoring .UCF file associated to our project. And we are actually able to customize our Decision 3. In this case, let's call the Decision 3 one as vibration monitoring monitoring. And let's add a couple of labels for each of the different states. There will be three states in this application reference, 0, 1, and 2. 0 represents no vibration. So let's type no vibration. 1 represents low vibration. And 2 represents high vibration. And let's save this configuration. In this way, you basically set your sensor tile dot box with the LSM6 DSOX to run the application example of the vibration analysis. And as you can see here in the settings, now my mobile app will recognize and will give me as an output the different values for no vibration, low vibration, and high vibration that I customized with my own labels directly through the mobile app interface. The next step is selecting the output that we are going to use. In this case, let's stream via Bluetooth. So streaming via Bluetooth, continue and then save application. Now it's time to save your application. Since it's your own custom design, you can actually give the any name that you want. So in my case, I'll call it example vibration monitoring. MLC based demo. Finish. And it will be listed as one of the available options now in the custom apps window. So let's click on the upload button here at the right hand side of the mobile app name. So example vibration monitoring, upload button. From there, your hardware, your mobile app will find your available hardware. In this case, my sensor tile box is listed here. I'll click on the play button and then the play button will basically ask me if I want to upload and overwrite the current application that is loaded into my sensor tile box. The answer is yes. Let's go ahead and then the app was loaded successfully. Hit OK. And now it's time to connect to our sensor tile box. So now that I'm connected to my sensor tile box, I can go here on the top left menu and select the machine learning core button. And as you can see, my decision tree will be ready to provide me outputs. So in this case, I'm not moving the board so no vibration state is listed. If I start moving my board slowly with a little bit of vibration, you will get a new state through the mobile app, which basically is low vibration. And then if I increase the vibration intensity, now it will change for high vibration. So this gives you an idea of how to get started and actually evaluate your machine learning core implementation. So just as a recap, the process to utilize the machine learning core examples in unical configuration file format that are available in our GitHub page directly through the STPLE sensor app and the sensor tile box consists of the following steps. The first one is entering the expert field of the app. From there, selecting the MLC virtual sensor as an input and by clicking on the settings button of the input, load the desired dot UCF file. By the way, you can not only load the examples that are available in our repository, but also utilize this functionality to test your own machine learning core design. And from there, once the expert design is finalized through the app, it's just a matter of uploading the new program into the sensor tile box and testing the results right away. It is also important to highlight that the data logging capability is always available when using the STBLE sensor app and the sensor tile box. And with that, I thank you very much for your time and attention and for more information, please visit st.com slash sensors.