5 Examples of Embedded Computers

When the latest GPU technology is combined with rugged computing hardware, Industrial AIOT applications can be realised right at the edge, and the applications in which AI is being utilised in modern enterprises is becoming increasingly innovative and interesting as system designers in the industrial and commercial spaces learn more about the potential of machine learning and neural networks.

The performance focus for Embedded Computers for Edge AI is generally centred around GPU capability. At the core of Artifical Intelligence lies deep learning — a process with an insatiable appetite for computing resources. The training phase in particular takes up the most resources, with intensive computation and analysis building the AI algorythms in real-time, and the introduction of an Embedded Computer with industrial grade GPU performance can help dramatically speed up the processing time for these deep learning exercises.

But why are GPUs faster at deep learning computational processes? It boils down to cores. CPUs have multiple complex cores (Multiple instructions, multiple data (MIMD)) which run processes sequentially with few threads at a time, whereas GPUs have multiple simple cores (Single instruction, multiple data (SIMD)). The difference between the two comes in the way threads are processed - MIMD cores process threads sequentially, with SIMD cores processing them in parallel. An analogy of this principal could be two boats - a speedboat (CPU) and a cargo ship (GPU). Both have the same cargo to transfer from the mainland to an island, with the cargo ship being able to take the entire payload in one go, and the speedboat having to make multiple trips. The cargo ship may move slower than the speedboat, but it completes the task in a fraction of the time.
 

• Embedded Computers for Edge AI
• MXM GPUs Bring Intelligent Video Analytics to the Edge

Machine Vision enables manufacturers to reduce operational costs and material waste, minimize the risk of recalls, and protect their brand reputations. Using high-speed cameras on the production line to ensure quality has become more and more common, and whether it’s ejecting milk bottles without lids from the line, inspecting incoming raw ingredients for foreign objects, or checking the 1200 LGA pins are all perfectly aligned, cameras are everywhere in modern manufacturing trying to catch issues before they become costly recalls.

Embedded computers for Machine Vision have highspeed I/O ports aplenty, whether it’s USB3.2 Gen2 10gbps with independent controllers, High-speed Gigabit and 2.5 Gigabit Ethernet ports,  GMSL camera interfaces or Gigabit PoE, they’ll often come with triggers and LED controls as well as traditional Serial I/O and Digital I/O to act when they detect the issue they are looking for.
 

• Embedded Computers for Machine Vision
• Nuvo-7006E
• eBOX671-521-FL

From simple passenger information systems to multi-screen video walls, Embedded PCs in the world of digital signage applications are perhaps the most varied in their graphical and performance requirements. When choosing an embedded computer for a digital signage or multimedia application the focus for system designers and OEMs most often falls on video output — from number and type of I/O, to graphical capabilties and video quality.

For example, with its slimline footprint Advantech's DS-082 is perfect to tuck behind an LCD screen, and its four HDMI outputs allow it to stream to up to four screens in high definition at any one time. The Nuvo-2700DS from Neousys takes that to the next level, streaming 4K UHD video to four displays at an impressive 60 FPS, making it ideal for rich, immersive multimedia applications.
 

• Embedded Computers for Digital Signage
• DS-082 - Ultra slim digital signage player
• Nuvo-2700DS - 4K at 60 FPS

An IoT Gateway is a locally deployed embedded computer that collects, stores, processes, analyses and acts on data provided by IoT devices. These computers can provide organisations with cost savings through reduced bandwidth usage by aggregating the data and transmitting a condensed dataset to the cloud.

When choosing the right embedded computer for your IoT project the scope of the application will dictate your choices: how many devices you’re connecting to, how you’re communicating with the computer, what protocols it needs to support, what software it needs to run. All of these things have an impact on how system designers and OEMs will choose the right embedded computer.

The EI-52 compact edge system from Advantech is perfect for intelligent IoT applications. It’s compatible with over 20 different data acquisition devices, and supports over 15 industrial IoT protocols including OPC-UA, Modbus, MQTT, REST and BACnet. The UC-8112-LX is more application-specific with just 2 Serial ports and a 32 bit single-core 1GHz ARM processor, but its compact size, flexible communication options, and wide power input make it a great choice for embedded data-acquisition applications.
 

• Embedded Computers for Industrial IoT

In-vehicle embedded computers, whether intended for the automotive, emergency services, marine, or railway rolling stock, will need to withstand the temperature variances, shock, and vibrations that are prevalent during mobile operation. Several different classification bodies offer certification to confirm the suitability of a device for in-vehicle applications, such as DNV for marine, EN 50155 for rolling stock, eMark for automotive. All of these certificates and standards define a set of tests and requirements that devices should meet to be suitable for installation in their intended vehicles.

When choosing an embedded computer for an application, system designers and OEMs will consider features that suit the intended application, such as GPS, GLONASS and Cellular support for fleet management, or PoE network ports for passenger safety and surveillance systems. Control systems require a lot of I/O, and passenger information systems will often require multiple display outputs. Other features, however, are shared, like lockable I/O connections, wide operating temperature, wide DC input, and remote ignition.
 

• In-vehicle Embedded PCs