The next phase of the ongoing development of the industrial sector is being referred to as Industry 5.0, or 5IR for short. Rather than launching a revolution, 5IR seeks to expand on the foundation established by its predecessor, Industry 4.0, which forever disrupted the industrial landscape by introducing big data, automation, cloud computing, artificial intelligence, M2M and the Internet of Things (IoT).
"Companies and whole industries are still in the midst of aligning their operations with Industry 4.0," said an Advantech spokesperson. "As they do, some are adopting the aims of 5IR. Chief among those aims is to foster greater collaboration between humans and Industry 4.0 technology, especially in tasks that require creativity, complex decision-making and emotional skills. In that way, companies that roll out 5IR get the best of both worlds: human ingenuity paired with automated efficiency to obtain a competitive advantage."
Being human-centric, 5IR has far-reaching implications for employees. 5IR brings them back onto the plant floor, a step that adds agility and flexibility to automated yet rigid manufacturing processes by enabling the workers to play an active role in decision making.
Although still in its early stages, some possible use cases for 5IR have been identified, including:
- Manufacturing: AI-enabled collaborative robots (cobots) and smart devices will play a larger role in 5IR than Industry 4.0. Like an apprentice, the new breed of cobot can visually observe complex human processes and learn to perform them so they can free workers from repetitive and monotonous tasks. Cobots can work safely alongside humans, which opens new prospects for manufacturing, instead of being gated off for safety like a traditional robot. Human and machines working together allows employees to focus on creating value for customers through customization and innovation. Smart devices can observe, analyze and comprehend visual data to recognize and distinguish differences due to specialized algorithms that automate visual understanding and pattern matching.
- Security: AI can uncover patterns of human behavior by analyzing video footage. This will help surveillance systems detect early actions that may predict a security incident or an access breach. Footage from retail surveillance cameras can also be a tool for gaining insights into customer needs, preferences and behaviors when combined with data analytics.
- Transportation: Integrating visual systems into vehicles and alongside roadways improves driver safety and environmental sustainability, while enhancing business distribution through reduced congestion and route optimization for logistics.
As these examples show, imaging technology is at the core of achieving many of the ambitious goals of 5IR. Machine vision in the 5IR age provide robots, cobots and machines with human-like "sight" combined with AI "insight."
Leveraging the capabilities of high-resolution digital cameras, lightning-fast networks, GPT's multimodal capabilities and AI-powered software will be necessary to take machine vision to this new level. Deep learning, a subset of AI, can teach a machine to contextualize the acquired visual data, so that it can mimic human cognition to make predictive decisions. Natural language processing also lets these same systems read and interpret information contained in visual data, such as reading labels on pharmaceutical packaging, as opposed to traditional rule-based machine vision approaches that necessitate large amounts of technical skill and considerable programming.
Transmission from Camera to PC
Although Industry 4.0 and Industry 5.0 represent two distinct phases in the evolution of industrial processes, they have one thing in common: a reliance on the transmission of bandwidth-intensive, real-time image data.
That being the case, the lynchpin to 5IR may lie in the space between the camera and the PC in the form of a frame grabber. The purpose of this computer accessory is to capture images from a camera and transmit those images to the host memory of the PC for processing.
There are a few different transmission standards for frame grabbers, with the most common being Camera Link (CL) over Ethernet cables, Camera Link over LVDS cabling and CoaXPress (CXP) over coaxial cables.
"When compared to nonframe grabber-based transmission methods like GigE Vision or USB3, CoaXPress frame grabbers allow faster, more dependable data transmission for an image system," said the spokesperson. "Since it was introduced in 2008, the CoaXPress interface has proven to be the ideal balance between system costs and growing requirements for faster speeds, extended length cables, heat dissipation and power delivery via power over CXP."
The latest version of CoaXPress, the CXP 2.1 interface, is particularly well-suited for 5IR imaging applications. It meets the need for speed by offering transfer data rates of up to 12.5 gigabits per second (Gbps) per link over a single coaxial cable or 50 Gbps over four cables when all four links are used for a single camera. Not only does CoaXPress carry image data over a single coax cable, but camera communication, control and power.
The latest add-on is CoaXPress Over Fiber (CoF). Industry experts foresee CoF serving as a viable pathway towards 100, 200 and even 400 Gbps, that is, speeds well beyond the capabilities of coaxial cables. In addition to speed, CoF is expected to achieve distances up to 80 km (about 49.71 mi.) in single-mode and 300 m (about 984.25') in multimode without the use of error-prone extenders.
CXP Combined with AI
Graphics processing units (GPUs) are a key lever to increasing productivity and competitiveness via AI-enabled technologies. Powered with thousands of processor cores and designed with a highly parallelable architecture, GPUs are accelerating high-performance computing workloads, deep learning and inference.
CoaXPress frame grabbers are increasingly being engineered for compatibility with the latest generation of GPUs to fast-track prototype system development and deployment of vision systems, robotics and sophisticated edge AI applications. This development merges the lightning-fast data rate speeds of CoaXPress 2.1 with the unprecedented computational capabilities and large shared memory for CPU and embedded GPU of devices, such as the NVIDIA Jetson Orin and the Advantech AIR-030 AI Inference System Box. Building complex machine vision and autonomous inspection applications is made possible by CXP/GPU solutions, which combine AI-accelerated image processing with interface support for numerous CoaXPress (CXP) cameras at up to 50 GB per second. It is also an ideal platform for prototyping end-to-end AI applications.
Industry 4.0 is being driven by advances in machine vision and AI-enabled technologies, which will also serve as the foundation of 5IR. Having the right vision hardware in place is vital to being successful in this journey. Machine vision, and specifically the CoaXPress interface, creates a more seamless interplay between humans and machines. In addition, it can help 5IR in reducing production waste, increasing sustainability and improving efficiency.
Authored by Donal Waide, Director of Business Development, Industrial Cloud & Video Group, Advantech
BitFlow, Inc.
400 West Cummings Park
Suite 5050
Woburn, MA 01801
781-932-2900
www.bitflow.com
Advantech
222 Rosewood Drive
Danvers, MA 01923
800-866-6008
www.advantech.com