 Welcome to Electronica 2018 where I'm delighted to say that I've been joined by Danny Lill Pau and Simon Actari to talk about artificial intelligence and great to see two exciting innovative companies working together here on the ST booth. How did that come about? Well Chris, I was demonstrating at Embedded Word 18, cube.ai tool that is the great tool to convert automatically any customer provided neural network into an efficient library that runs on any STM32 MCU. I was demonstrating that functionalities to different people. I received the visit from Simon Actari, Leonard and Bauer. We started to chat together about cube.ai, demonstrating things, sharing the interest and well, we were on the same page and the story started. Right, great start to a happy marriage then Simon. Yes, when I met Danilo at Embedded Word I immediately recognized the potential, the tool that's to offer enabling us to implement artificial intelligence into our sensor solutions and of course I'm very happy to be here with the joint demonstrator here at Electronica Munich. So have you guys been in lockdown working on experiments for months then? Yeah, we were in submarine mode for a lot of months since March. We did a lot of interaction. Simon was working hard on collecting data, architecting networks. I was used to convert those networks, many of those using cube.ai. At the point I was also bringing by motivation the tool and the board's home to convert overnight the networks and quickly providing support to Simon. Now Simon, you're going to talk me through the demo. Just put it into context for me. Yes, so at Leonard and Bauer we are developing and producing sensor solutions for mobility industry with focus on railway technology and also machinery tools for method processing. For example, we are providing solutions for train maintenance. So we have for example a bogie which after a certain amount of time experience some material fatigue and for safety reasons we have to check regularly the state of the ball bearings. Up until now we are doing these things in a complete manual approach which means that we are analyzing the sensor output by hand and setting thresholds accordingly. But with HAI we are creating a solution that can improve our efficiency enormously by just recording the sensor data, building up a database that represents the machine conditions, training the neural network and porting it to the microcontroller. So this is it in action, talk me through it. Okay, on the left side you can see attached sensor tile with a 3D accelerometer measuring the vibrations of the whole system. Now with the small hammer on the right we can induce more vibrations to the system and also what is very important is that we have the neural network, the artificial intelligence directly on the microcontroller next to the sensor. Now if I switch on the hammer we see that the color of the LEDs is changing because the neural network recognizes the vibrations and classifies it accordingly. So now I can induce even more vibrations and we see the LEDs changing in color even faster than if we take a look here at the PC because we have to transmit and send the data to the PC where it's then visualized. And I understand there's something called edge AI involved in this? Yes, so artificial intelligence for a lot of people is that you send a huge amount of data to the cloud where you have huge amount of processing capabilities, have huge neural networks analyzing your data. But HAI means that you have the intelligence directly at the sensor level with low power consumption and very low latency which is very important for our applications. Great, right Danilo bring cube AI into this. At Embedded World 18 we were demonstrating the main functionalities of cube AI. Since then we did a lot of technical work to integrate those functionalities in the tool that is called STM32 cube mix. Cube mix is the tool that all the customers across the world use to map any software on the STM32. So we fuse together the two functionalities, cube AI into cube mix in order to start sampling the dead capability to some customer, the alpha customer starting from October 18. And general release, Q1 maybe next year? Yes, this is going to happen Q1 2019. The mass market will start receiving through ST.com the tool and so we'll be free to use it. Well this is great to see this partnership working so well together. It just shows you that events like this can bring together really innovative companies in that wonderful sort of synergy and partnership. It's been great talking to you, best of luck. Thank you.