 Hello! In this short video, we will describe the different functionalities of the Sensortile.box wireless IoT and wearable sensor development kit. This is the third video of the series, and we will look into the so-called Pro Mode of the Sensortile.box. The Pro Mode is enabled by opening the blue box and connecting the Sensortile.box to the STM32 LINK V2 or LINK V3 programmer. In case of LINK V3, a voltage level shifter is required. This allows users to configure the Sensortile.box to best fit their application. Download the zip package containing the code for the Pro Mode examples on ST.com. Due to the package organization, it is recommended to unzip it in a short directory path, so the closest to C colon backslash as possible for Windows users. The Pro Mode functional pack contains seven different projects and all the required drivers and middleware for using the Sensortile.box board. Since this package is continuously updated, please check online to find its latest version. The driver directory displays all the drivers for each component included in Sensortile.box and the board support driver for.box. The middleware directory includes four middleware packages, Bluetooth, the FATFS file system, the free RTOS real-time operating system for multithreads examples, and the USB-D device library. There are seven different projects contained in the Pro Mode function pack. Four applications are available. How to make sure secure Bluetooth pairing and firmware over-the-air update, BLE FOTA application. BLE Sensors, the simplest way to send sensors data to one application using Bluetooth. BLE Low Power, how to use free RTOS and low power techniques to send the same information. BLE MLC, one example on how to program and use the Machine Learning Core in the LSMSDSOXIMU. You can also find one application on how to save sensor data and audio capture by microphone on SD card or Datalog. Finally, two examples are included. Bootloader, it is the bootloader necessary to implement firmware over-the-air update to BLE FOTA. Datalog Extended, to stream sensors data to a connected PC using the USB-D and compatible with Unicleo GUI. The BLE Power and Datalog applications use the free RTOS real-time operating system in order to reduce the power consumption and enable maximum speed. The BLE FOTA application and the bootloader example must be active at the same time on the board and they are placed on different flash regions. The current BLE FOTA application, old, receives the new version, the FOTA application with the BLE. It saves the new BLE FOTA version on the third flash memory region. The dot box will reboot to allow the bootloader to replace the old application with the new one just received. As example and practical exercise, we will now see how to test the Human Activity Recognition Algorithm computed by the Machine Learning Core running on LSM-6D SOX. Let's get started. The objectives of this exercise are to show how to enter in DFU mode. Load pre-compiled machine learning firmware. Test the Machine Learning Core or MLC. Using the MLC design flow. Three steps are required to change the firmware running in sensor tile dot box. All the steps are described in the following slides. Enter in DFU mode using the STBLE Sensor application following the steps described here. First, open the debug console. Once the debug console is opened, initiate the DFU in capital letters command to reboot in DFU mode. Run the PC STM32 CUBE PROG program and follow the steps to connect to the sensor tile dot box. The device in DFU mode gives access to its internal flash memory. At this point, we need to select a race and programming to move on. Let's perform a full chip erase and confirm once prompted. Select the binary you want to upload, i.e. BLE-MLC dot bin. See if you want the verification option to be active and press start program. Once the uploaded binary has been verified OK, press OK and disconnect the USB cable. Now let's connect the BLE Sensor mobile application to the sensor tile dot box. The app will show the gyroscope and accelerometer data in real time. The second screen displays the icon matching the activity recognized and loaded in the MLC. Since walking and running are related to some mainly vertical movements at a regular pace, by gently shaking the sensor tile dot box, the walking movement will be recognized or running depending on shake frequency. If properly stimulated, the matching icon will turn blue showing the MLC has detected that class. Now let's go into further details. The BLE-MLC application included in the Pro mode of the FPSSNS-STBOX-1 functional pack uses the machine learning core embedded in the LSM-60SOX-IMU to identify the activity, stationary, walking, running and so forth. The MLC can be programmed using the Unico GUI for collecting data, testing and creating the right program for the machine learning core. Unico GUI is a graphical interface to elevate ST sensors. It can be used in many different ways, connected or disconnected from an evaluation board. In this case, it is used to generate the MLC programming sequence together with another freeware tool, WECA, provided by the University of New Zealand. First steps are executed in the Unico GUI. We need to start from the data collection. This is probably the most critical step since the performance of the recognition algorithm largely depends on the quality of the collected data. The classes to be recognized are defined here. After this step, an external statistical analysis program is invoked and the execution is passed on to it. This tool analyzes the collected data, the classes, and helps define the most useful data features for the classification. Once the statistical analysis is completed, a result file is saved on the PC. Alternative tools can be used to generate the decision tree instead of WECA, RapidMiner, MATLAB or the Python library. Please use the link shown in the video to download the one you prefer. Unico GUI, by importing the output from WECA or other tools, can save the .h file with the corresponding program necessary for the machine learning core to operate. The .h file is a sequence of registers inside the LSM-6D-S-O-X-I-M-U. A simple C program can execute this register loading when devices are powered up. Inside the BLE-MLC application provided inside the Pro Mode of the FPSNS-STBOX-1 function pack, there are two different programs to understand the machine learning core. One of them is the example just carried out, implementing human body activity recognition. The other one is for equipment vibration monitoring. Additional MLC examples for machine learning core are available on ST's official GitHub repository. Thanks for following this video presentation. The Pro Mode depends on users and applications. Only one example was presented here, but many more are possible. We hope this presentation will help you get started on the Censortile.box. Don't hesitate to join our online community to share insights or ask questions.