 Hello and welcome to this series of Getting Started with STM32Cube.AI and Motion Sensing on STM32L4 IoT node. The video application node series is divided in three sections. In the first section we will go through an introduction of the STM32Cube.AI, the Human Activity Recognition and the prerequisite for the end zone. In the second section we will create the project using STM32Cube IDE and we will configure the peripherals needed to acquire the sensor. In the last section we will add the neural network to process the accelerometer data and obtain the Human Activity Recognition using STM32Cube.AI, a brief introduction of ST microelectronics and our AI solution. As one of the world's largest semiconductor companies, ST has been actively involved in AI research topics. Thanks to the new set of AI solutions from ST, you can now have the possibility to map and run pre-trained artificial neural network on the broad STM32 microcontroller portfolio. You can use the power of deep learning to announce signal processing performance and increase the productivity in your STM32 application. There are five key steps behind running neural network on MCUs. First capture data, then clean and label collected data and build the network topology. Train the neural network with the data and convert the trained model into optimized code that runs on STM32 microcontrollers. And finally, run the neural network on the device. Here we can see the last three steps. Normally the training of neural network is done by data scientists or machine learning engineers, while the implementation of MCU is done by embedded engineers. STM32Cube.AI is a convenient and efficient conversion tool in between. It is the bridge linking these two steps. Besides, STM32Cube.AI is also an extensive toolbox to support easy creation of AI applications. We have examples, community support, various training, end zone workshop, as well MOOCs, and partner video from our partner program. In this application note series, we will learn how to create a motion sensing application to recognize human activities using machine learning on an STM32 microcontroller. The model used classifies activities such as stationary, walking, or running, from accelerometer data provided by the LSM6DSL sensor. We will be creating a human activity recognition application by learning how to read motion sensing data, how to generate neural network code for STM32 using XCube.AI, how to input sensor data into a neural network code, and how to output the interference results. Here we can see a high level block diagram of the application. We have a first initialization phase with the configuration of the peripherals, like I2C and UART. Then we have an infinite loop where we acquire the accelerometer data, and when we have enough data we call the neural network interference. The neural network will output the human activity recognition, such as stationary, walking, or running. In this example, we are using a simple neural network model. We are more focused on the flow with STM32Cube.AI and STM32Cube.AI. Before getting started, here is the list of software and hardware we will need. We will use STM32Cube.AI version 1.5.0 or later, and within Cube.AI you will be able to install XCube.AI version 5.2.0 or later for the neural network conversion tool and neural network runtime library, and XCube.MEMS 1 version 8.2.0 or later for the motion sensor component drivers. We are going to need a serial terminal application, such as TerraTerm, and for the hardware we will use the STM32L4 discovery kit IoT node, that is the B-L475E-IoT01A. Here you can find some useful links. The first one is the wiki page associated with this video application node. You can use this link to copy paste the code described in the next videos. You can use the second link to download STM32Cube.AI. In the third link you can find the pre-trained Keras model that we will integrate in our application. You will also find data sets used to train the model and a Python notebook to train your own model. Finally, a link to the FPAI sensing one, the STM32Cube function pack for ultra-low-power IoT node with artificial intelligence applications based on audio and motion sensing.