Lattice Blog


mVision Debuts at Embedded World 2020

Embedded World 2019
Posted 03/12/2020 by Dirk Seidel

Posted in

I attended Embedded World in Nuremberg, Germany recently and wanted to share some thoughts about the trends and technologies that received the most attention at the show this year. Like last year, most of my conversations with customers and developers at the show were around supporting AI in embedded applications for the automotive and industrial markets. But unlike last year, customers have more fully formed roadmaps for implementing AI support in their products. And most of those roadmaps start with support for embedded vision.

Let me explain. If you’re going to add AI to a system, you need sensor data to inform your AI analytics. Image sensors are the dominant sensor technology in use today, so in their pursuit of AI support, many embedded device developers are first looking to add support for embedded vision. This creates a design challenge caused by the fact that many of these embedded vision systems will be operating on the Edge of the network. Because they’re often implemented in challenging usage environments (hot factory floors or in compact spaces with little ventilation) Edge devices need to carefully manage their thermal performance, which has many developers looking for low power, small form factor design solutions to quickly and easily enable embedded vision.

This was the perfect context for the debut of our new Lattice mVision™ solutions stack for low power embedded vision. mVision is latest addition to our growing portfolio of complete system solutions designed to provide developers with the hardware and software they need to quickly and easily implement popular emerging technologies (like AI and embedded vision) on Edge devices. Our first solutions stack, sensAI, received great acclaim in the industry; it’s received five industry awards since it launched. And it appears our mVision solutions stack is following in the same footprints as sensAI; it won a “Best of Show” award from Open Systems Media.

The mVision solutions stack includes the Video interface Platform (VIP) (a set of modular hardware development boards), design software, embedded vision IP portfolio, and reference designs and demos; everything needed to implement sensor bridging, sensor aggregation, and image processing applications. The stack features support for Lattice FPGAs, including CrossLink™, ECP5™ and our new CrossLink-NX™, based on the industry’s only 28 nm FD-SOI-based platform for low power FPGA development, the Lattice Nexus™ platform.

Lattice regularly presents papers at the technical conference portion of Embedded World, and this year it was my turn to represent the company. My presentation covered how to implement a very low power embedded vision system to act as a human-machine interface (HMI) solution with support for simple gesture detection. The design, based on our very low power iCE40 UltraPlus™ FPGA, can be modified to perform a wide range gesture-based HMI functions in all sorts of possible applications, from using a hand gesture to turn office equipment and displays on or off, or to adjust thermostats and other IoT- and IIoT-connected control appliances in a smart home or factory.

A special shout out to everyone who visited Lattice Semiconductor at Embedded World; we know travel was particularly challenging this year. We hope you found our product and application demos informative and inspiring. We’ll see you again in 2021!

And for some more insight into the embedded vision market and how FPGAs enable embedded vision applications, read the latest blog from tech analyst Bob O’Donnell.