Asus IoT launches ultra-compact computer for AI inference in manufacturing

Hand holding icon iot of wood cube block
Sheila Zabeu -

July 10, 2023

Asus IoT, the arm of Asus providing Artificial Intelligence of Things (AIoT) solutions, has launched an ultra-compact computer for Artificial Intelligence (AI) inference at the edge. The new PE1100N uses the NVidia Jetson Orin platform and features a fanless design to ensure quiet operation and advanced connectivity features.

According to Asus IoT, the PE1100N is suitable for use in industrial environments, smart cities and transport systems as it offers a variety of Artificial Intelligence (AI) applications on its easily extendable edges, such as counting, surveillance, traffic analysis and people tracking. The compact size and anti-vibration design make the PE1100N suitable for smart manufacturing solutions such as automated guided vehicles, autonomous mobile robots, AI-powered automated optical inspection and robotics applications.

Asus IoT has developed the PE1100N to directly serve the growing smart manufacturing market, in close collaboration with NVidia. Integrated with the NVidia Jetson Orin platform for AI at the edge, the PE1100N can perform up to 100 trillion operations per second (TOPS), with low power consumption (10-25W), in deep learning and computer vision tasks. According to Asus IoT, this represents five times higher performance than the previous generation. It also achieves up to 18 times higher frames-per-watt efficiency than traditional x86-based platforms.

In addition, the PE1100N runs AI and machine learning models, using standard SDK and library tools with minimal code changes. Because of this, the Jetson Orin platform can be adopted in an easy way, making it easier to create and deploy projects. The PE1100N can also work with the NVidia Isaac Robot Operating System (ROS) SDK.

The PE1100N has a robust design that includes a metal heatsink and an aluminium chassis that enables efficient and quiet cooling and ensures reliable operational stability over a wide temperature range (from -20° to 50°C), according to Asus IoT.

It comes equipped with an extensive collection of I/O ports for various industrial purposes (LAN, DIO, COM, USB 3.2 and Type-A and, on some models, CAN bus). It also features a Micro-USB debug port for easy system maintenance. An optional WiFi and Bluetooth module ensures wireless and cloud connectivity, while the M.2 B switch can accommodate a module for 4G/5G cellular connection. In addition, compatibility with GPS solutions enables accurate tracking and data recording in smart transport solutions.

The PE1100N is available in two distinct types, with NVidia Jetson Orin NX or Jetson Orin Nano, to meet different AI performance requirements. There is also an on-board configuration to meet customised requests.

More than AIoT, now IoT and Generative AI

Some say AIoT works better as a concept than as a buzzword. The term AIoT started appearing in social media posts in 2019 and grew rapidly over the next two years, but levelled off in 2022, perhaps overshadowed by the emergence of the new buzzword, generative AI.

And why not think of IoT combined with generative AI? A nomenclature for this potential duo has not yet been coined, but there are already examples of applications that even go beyond the famous ChatGPT. These use cases have been collated in a detailed scenario report by IoT Analytics. Here are some of them:

1. Code generation for IoT: Models proposed by generative AI can help in the creation of IoT applications, together with many development environments (IDEs) that are also already compatible with this class of Artificial Intelligence.

2. Robot control: Generative AI can generate control logic and commands for robots by capturing motion data from animals or people. So, instead of programming movements for each limb of robots, generative AI models can, for example, be used to generate movements of individual parts so that robots take more complex and interconnected steps. In addition, these models can help robots understand the environments in which they operate to achieve certain goals.

3. Industrial IoT product design: Generative AI can automate the design process by generating a large number of options capable of meeting specific requirements, performance criteria and constraints. By combining generative AI and CAD, designers and engineers can create more efficient and innovative designs.

4. Social IoT devices: Generative AI can make communication “more social” by allowing devices to answer complex questions or users to communicate with devices to change their settings. It can also allow devices to generate more responses.