Bench Talk

A Practical AI Edge Solution for Advancing Industry 5.0

New Tech Tuesdays

Join Rudy Ramos for a weekly look at all things interesting, new, and noteworthy for design engineers.

In just a few years since OpenAI released ChatGPT, the generative artificial intelligence (AI) chatbot has permeated every area of the global economy. AI technology, formerly limited to educational institutions and major enterprises operating in secret, has spread to every part of society, including our homes, businesses, and industries.

In this week’s New Tech Tuesday, we examine how tiny single board computers (SBCs) like the Raspberry Pi are a viable solution for bringing AI to the industrial automation edge.  

Industrial Automation in the Digital Age

For decades, many industrial and manufacturing operations relied on rigid, low-variance machinery with little to no intelligence. These operations employed time-tested maintenance schedules and technical support teams to keep the machines working and the operation moving.

To thrive in the digital age, many companies in the industrial sector are transitioning to Industry 5.0, focusing on industrial automation that integrates AI and machine learning (ML) with human-centric values and emerging technologies like edge computing. At the same time, industrial Internet of Things (IIoT)-enabled systems can operate faster than manual labor by automating repetitive tasks. IIoT further reduces downtime and optimizes processes by identifying equipment failures before they occur and making data-driven decisions based on analysis of machine performance, inventory levels, and more.

However, all this requires distributed nodes such as cameras, robots, sensors, programable logic controllers (PLCs), and gateways to send the raw data to IoT platforms for processing, which typically involves sending the data to a centralized location or data center commonly referred to as the cloud.

Bring AI to the Industrial Automation Edge

Edge computing offers industrial automation several benefits over the cloud. It provides real-time data processing locally, which is ideal for manufacturing operations that require quick responses. Analyzing data close to the network’s edge delivers faster results and responses by reducing the latency for data to travel to and from a centralized data center for processing. Edge computing also reduces the need for expensive cloud computing resources and bandwidth requirements. Lastly, edge applications can continue to function even if cloud connectivity is impaired.

Of course, some challenges are associated with edge computing, primarily regarding security, scalability, and computational resources. Due to their dispersed locations, edge devices—and their data—are more vulnerable to bad actors and the environment. The data being processed at the edge may not have the same level of physical or cyber protection safeguarding it from malware and other cyberattacks. Edge devices frequently lack sufficient computing power or networking capabilities, complicating software and firmware updates.

The Newest Products for Your Newest Designs^®

This week’s New Tech Tuesday highlights AI-enabled edge computing with cost-effective hardware from Raspberry Pi that tackles security, scalability, and computational constraints.

The Raspberry Pi is one of the most popular and extremely versatile SBCs. The latest iteration, the Raspberry Pi 5, features a more powerful processor, larger memory, and greater input/output flexibility compared to other SBCs in the market. This makes it an excellent choice for more advanced projects, including robotics, media centers, IoT applications, and now advanced AI capabilities with the introduction of the Raspberry Pi Hailo-8L AI Kit (Figure 1).

Figure 1: The Hailo-8L AI kit is a pre-assembled, ready-to-use solution designed for edge AI tasks. It integrates an M.2-format Hailo-8L AI accelerator with the Raspberry Pi M.2 HAT+. (Source: Mouser Electronics)

The Hailo-8L module is based on the entry-level, 13 tera-operations per second (TOPS) Hailo-8L AI processor,^[1] capable of delivering exceptional AI performance for edge devices. The Hailo-8L features a distinctive, scalable dataflow architecture that leverages the core strengths of neural networks, allowing edge devices to run deep learning applications more efficiently and effectively than traditional solutions. The Hailo-8L AI kit improves industrial automation with high-resolution edge defect detection using multiple cameras running simultaneous learning tasks and is available to order from Mouser Electronics.

Tuesday’s Takeaway

AI has rapidly permeated every sector of the global economy, from automotive and security to retail and personal computing. AI is transforming how we use machines to perceive and analyze our environment in real time. Now, AI is poised to transform industrial automation by making AI at the edge more efficient, secure, and affordable using single board computers.

Sources

[1] https://hailo.ai/products/ai-accelerators/hailo-8l-ai-accelerator-for-ai-light-applications/#hailo8l-overview