NVIDIA Releases Jetson AGX Xavier Industrial Module, Engineered to Bring AI to Harsh, Safety-Critical Environments

In recent years, a wide range of industries such as manufacturing, agriculture, construction, energy are adopting robotics and automation technologies for multiple tasks, including inspection and maintenance, that are challenging for a human to accomplish in potentially dangerous environments. However, most companies struggle to incorporate AI in such demanding applications.

NVIDIA has launched a new Jetson AGX Xavier Industrial module that will enable developers to build advanced, AI-enabled ruggedized systems to deploy AI in the most demanding environments. 

This module finds application in various sectors, including aerospace, defense, construction, agriculture, delivery, inspection, logistics, inventory management, and healthcare. It is built for intelligent video analytics, optical inspection, robotics, Computer vision (CV), among others, to access and monitor the site, and inspect worker and site safety in hazardous or harsh environments.   

It comprises of a compact and power-efficient design that delivers up to 30 trillion operations per second of AI performance. Furthermore, the components used to build this module are brutally tested based on industrial standards and including new functional safety capabilities. As a result, these components are able to withstand severe shock and vibration and even extreme temperature.

Being pin-, software- and form-factor compatible with the existing Jetson AGX Xavier module makes its gradation significantly easy.

Source: https://blogs.nvidia.com/blog/2021/06/15/jetson-agx-xavier-industrial-use-ai/

Most Demanding Industrial Use Cases

The new module is capable of providing real-time insights for industrial applications. For example, in Oil and Gas facilities, it can streamline anomalies and predict any potential failures. It is able to draw these insights by monitoring pipelines, valves, equipment, maintenance work, etc. This feature is useful where safety, maintenance, and equipment compliance changes with environmental conditions. 

In the Manufacturing industry, where tens of thousands of products are produced every day, humans will not be enough to inspect millions of welded points. Employing AI and CV on the Jetson AGX Xavier Industrial allows the module to analyze process and quality data directly from weld guns. Automating such inspection tasks not only reduces inspection time but improves quality prediction as human inspection is subjected to errors.  

A similar AI application was presented at GTC21, where AI is employed to detect anomalies in die-casting processes. This AI-empowered detection system allowed early prediction of product quality. Therefore it avoids further process costs by up to 30 percent and decreasing defect rates by up to 40 percent. These systems not only provide more efficient and safer solutions to the process but also improves productivity and return on investment. 

Components and Features

The new module is developed by combining the supercomputing capabilities of the Jetson AGX Xavier system-on-module with new features so as to deploy AI in harsh environments. These added features include:

  • Error correction codes
  • Single error correction
  • Double error detection
  • Parity protection that delivers internal RAM resilience
  • Address and data bus error detection and correction
  • IP resiliency in industrial applications. 

The new module has functional safety capabilities and thus is suitable for safety-certified industrial-grade products. The safety abilities are supervised by the Safety Cluster Engine (SCE) that contains dual Arm Cortex-R5 processors. These processors are used for integrated fault-detection mechanisms, lock-step subsystems, and empower built-in-system tests. In addition, the Cortex-R5 can be used for safety and error correction functions as it is in an always-on domain.

Additionally, security features such as hardware-verified secure boot, hardware-accelerated cryptography, support for encrypted storage and memory are added to protect customer software. 

Jetson AGX Xavier Industrial is powered by the NVIDIA CUDA-X accelerated computing stack and JetPack SDK support. It is an entirely software-defined platform that enables cloud-native capabilities. This will allow developers to use cloud-native technologies to maintain and update ruggedized and safety-critical systems in the field for management. 

Jetson AGX Xavier Industrial module is available on the NVIDIA store to order in late July. Developers can use Jetson AGX Xavier Developer Kit and download JetPack to build and deploy deep learning and AI training and inference systems. 

Source: https://blogs.nvidia.com/blog/2021/06/15/jetson-agx-xavier-industrial-use-ai/