 Welcome to this demonstration by ST, our floor type detection. We have here a vacuum robot like there are millions in our households throughout the country and throughout the world. And ST is helping develop those new technologies that are going to empower those devices to do more on existing microcontroller platforms. In that case we are using a time of flight sensor L553 by ST to detect which floor type we are rolling on. We have about 50 different materials that have been recorded and we've created a machine learning neural network very easily. Fed it through QBI to an F401 standard STM32 microcontroller and thanks to QBI the total footprint of this solution is 68 kilobyte of flash, 1.6 kilobyte of RAM. This fits on any one of your solutions. You can implement and create your own machine learning algorithm, your own neural networks in your preferred framework and implement them on your STM32 boards. Be it on an existing board already implemented in device if you have the overhead or on a dedicated board to run full time the analysis. In the case of this demonstration we can see on the robot itself the different color of lights that are going to be appearing depending on the floor type it detects. So in that case we have 50 different materials that are being detected and classified into two main classes, soft and hard. This is relevant because this will dictate the behavior of the robot and what it's going to be doing to adapt to the floor it's on. In fernstheim on nefo one hour is 7 milliseconds. This means that you can check, recheck and re-recheck before rolling on anything. You can look ahead, you can look underneath the robot, whatever makes the most sense for your application. We have something that is light and that is adaptable to either another vacuum robot or any other floor type detection solution that you would like to implement. You can find more details on this demonstration and the tools used to build it in the link in the description below. Thank you.