Industrial 4.0 Concept Map

The emergence of Industry 4.0 is gradually changing industrial production, based on the digitization of physical processes and the introduction of cyber-physical production systems in manufacturing workshops. A large number of Industry 4.0 applications are based on the integration of Artificial Intelligence (AI) and the Internet of Things (IoT). Today, Edge AI is one of the most popular implementations of this AI-IoT combination, bringing AI capabilities closer to the network edge and supporting real-time data processing and analysis on edge devices.

Edge AI enhances the performance, timeliness, and safety of various use cases in manufacturing, including predictive maintenance, real-time quality control, and supply chain optimization. These use cases provide tangible benefits to key stakeholders such as manufacturers, maintenance teams, quality control departments, supply chain managers, factory operators, and workers.

Implementing Predictive Maintenance with Edge AI

Predictive maintenance is crucial for ensuring continuous machine operation and preventing costly downtime. Today, most predictive maintenance applications deploy machine learning models in cloud infrastructure to predict asset parameters such as Remaining Useful Life (RUL). Edge AI is essential for enhancing these use cases to minimize downtime and optimize maintenance. In this direction, real-time computations of Remaining Useful Life (RUL) and End of Life (EOL) can provide insights into the health and performance of machinery and other industrial assets.

For example, Edge AI can detect anomalies and deviations from expected behavior in a very short time (i.e., almost in real-time). Based on signals regarding these deviations, maintenance teams can make informed decisions in a timely manner, proactively address potential failures, and optimize maintenance schedules for critical assets. Edge AI also reduces the attack surface for predictive maintenance and smart asset management systems; ML models used for RUL calculations can be executed within edge clusters or devices rather than in cloud data centers. This is fundamental for enhancing the security of predictive maintenance and smart asset management solutions.

Example of CADReady Robotics using Edge AI for predictive maintenance on Forge Edge robots (Source: Dominic Pajak)

Innovating Predictive Maintenance with Open Source Edge AI Solutions Using Arduino

The fusion of IoT and Edge AI is crucial for addressing the limitations of traditional cloud systems, as it enables real-time decision-making closer to the data generation source — which is particularly important in manufacturing, where instantaneous responses can prevent costly machine failures. Today, powerful products like Arduino Pro’s Opta and Portenta Machine Control are the most robust and user-friendly catalysts for embracing Edge AI and revolutionizing predictive maintenance in small and large factories. Opta is an innovative micro PLC that is quick and easy to use, gaining significant traction in the industry due to its support for PLC standard languages in addition to the Arduino programming experience.

Opta microPLC (Source: Arduino)

Portenta Machine Control is a fully centralized low-power industrial control unit that many businesses have successfully chosen to provide IIoT capabilities for both new and legacy machines, and its modular design allows for easy adaptation to a wide range of applications.

Portenta Machine Control (Source: Arduino)

For example, leading capping machine supplier AROL pairs Portenta Machine Control with Arduino Pro’s Nicla Sense ME module to integrate monitoring and predictive maintenance capabilities into the equipment they sell, significantly enhancing the value they provide to customers through efficient data processing and wireless communication.

Nicla Sense ME (Source: Arduino)

The Spanish engineering company Engapplic chose Arduino’s Portenta Machine Control to monitor the efficiency of an air compressor for a demanding client in the automotive sector, enabling timely detection and even prediction of any anomalies. The result is a cost-effective, future-proof PoC that reduces downtime and saves energy.

While predictive maintenance is an important reason to add Edge AI capabilities to machines, it is not the only one. By integrating the brain of Portenta Machine Control into installed bases (e.g., professional kitchen appliances or office printers and copiers), manufacturers can not only improve user experience and customer service but also create new business models, such as usage-based leasing contracts.

Arduino’s commitment to open-source principles provides additional advantages for industrial customers investing in advanced predictive maintenance solutions:

– No vendor lock-in: The company’s open-source approach ensures you can flexibly program, customize, and extend its solutions independently, avoiding vendor lock-in.

– Shorter learning curve: The user-friendly characteristics of Arduino products enable engineers to quickly grasp and implement solutions, even with limited programming backgrounds, facilitating efficient adoption by existing teams.

– Complete customization: With company support, you have free access to upgrade and modify Arduino solutions, ensuring seamless processes. This flexibility is especially valuable for companies seeking innovation and adaptability to changing demands.

Combining IoT and Edge AI for predictive maintenance in manufacturing is a transformative journey — Arduino’s powerful products and open-source approach can help navigate easily and innovate endlessly.

Real-Time Quality Control

Maintaining high product quality is crucial for manufacturers to meet customer expectations and comply with industry standards. Edge AI leverages inference on devices to detect anomalies and defects, enabling real-time quality control. By deploying AI models directly on edge devices, manufacturers can analyze sensor data in real-time, identify anomalies or defects, and take immediate action to correct issues.

This approach significantly reduces delays in the quality control process. Thus, it enables manufacturers to timely identify and address quality-related issues, reducing waste, promoting sustainability, and ensuring that products in the market are of high quality.

Case Study: Renesas Electronics

Enhancing Quality Control with Renesas Electronics’ Edge AI Tools

Edge AI has gained attention in the quality control field for its ability to run AI/ML algorithms on edge devices, enabling scalable real-time solutions. This approach is compatible with various MCU and MPU edge devices from Renesas Electronics, making it well-suited for large-scale manufacturing environments. It offers advantages such as real-time analytics, reduced latency, improved data security, and cost savings without the need for extensive data storage and communication infrastructure. Additionally, Edge AI systems can adapt to dynamic manufacturing processes effectively.

For example, in a typical welding production line, multiple robots perform various automated tasks. Here, end-of-line testing is critical for quality control, especially for components specifically manufactured. The challenge lies in detecting porosity and burn-through in the welding process using traditional methods. Therefore, there is a need to reduce reliance on manual inspection and implement low-cost solutions with real-time anomaly detection and high accuracy. Strict quality control ensures that no defective parts are misclassified as good parts, thus avoiding costly recalls by automotive OEMs (although there is a certain tolerance for misclassifying good parts as bad, which are discarded before shipment).

Reality AI Tools® (Source: Renesas Electronics)

Thus, the core task is to identify, collect, and analyze datasets, and then unify performance (inference), footprint, and accuracy in the most suitable AI/ML model. Selecting sensors and their positions within the system is crucial to ensure that the datasets are reliable and of high quality. Renesas Reality AI Tools® provide users with automated exploration of sensor data and perform analysis to find the best sensors (or sensor combinations) and optimal placements. With AI-driven capabilities in Reality AI Tools® Discovery, users can perform advanced automated exploration of sensor data and ultimately generate optimized AI/ML models based on the explored datasets.

To accelerate the development and deployment of dedicated Edge AI/ML solutions, Reality AI Tools® use a machine learning-guided process to explore data and create a set of custom transformations (feature space) to define anomalies or maximize separation of classes with respect to target variables. Users can inspect the feature space and generate time-frequency heat maps to show the structures most important for model accuracy. Achieving the highest accuracy and meeting a 0% false positive rate is the result of setting high standards for manufacturing efficiency and productivity.

The scalable product portfolio includes 16-bit, 32-bit, and 64-bit MCUs and MPUs, along with a rich ecosystem and development infrastructure for embedded systems and Edge AI/ML solutions, making it a perfect fit for system solution packages. It is designed for rapid prototyping, evaluation, development, and deployment of your next embedded Edge AI/ML solution.

Renesas Electronics Product Portfolio (Image Source: Renesas)

Facilitating Supply Chain Optimization

Efficient inventory management and logistics are crucial for maintaining streamlined manufacturing processes. Manufacturers continually seek to identify supply chain challenges to take remedial actions and implement optimizations. With Edge AI, they have the opportunity to identify and confront issues faster than ever. In particular, Edge AI can revolutionize supply chain optimization by timely detecting real-time events. Specifically, the collection and analysis of data from various sensors and devices enable Edge AI systems to gain real-time insights into inventory levels, demand fluctuations, and transportation status. Manufacturers can leverage these insights to make data-driven decisions, such as optimizing inventory levels, adjusting production schedules, or rescheduling shipping routes. This can significantly enhance overall supply chain efficiency, reduce costs, and ensure timely delivery of products.

Case Study: OKdo

Transforming Supply Chain Organizations: Practical Applications of OKdo’s Edge AI Solutions

In supply chain organizations, precision, efficiency, and safety are critical, necessitating the integration of cutting-edge technologies. OKdo offers a variety of AI-embedded products that drive faster and better decision-making through edge data collection and analysis.

Micro Code Reader from Useful Sensors: A Huge Leap in Inventory Control

The Tiny Code Reader from Useful Sensors is a groundbreaking supply chain solution equipped with an onboard processor, camera, and AI accelerator. The Tiny Code Reader is compact and features Qwiic connectivity, making it an ideal choice for seamless integration into various systems, simplifying inventory management. This micro product is about the size of a coin, maximizing space efficiency in warehouses and appealing to engineers who value flexible layouts. Its cost-effective design does not compromise technical prowess, providing comprehensive developer guides with sample code for popular systems. Imagine a warehouse with unparalleled accuracy. The Tiny Code Reader reduces errors and transforms inventory management for engineers at the forefront of supply chain innovation.

ROCK 5 AIO from OStream: Enhancing Predictive Maintenance and Safety Accuracy

Predictive maintenance is crucial for preventing downtime and ensuring the smooth flow of supply chains. The ROCK 5 AIO integrates 3 TOPS AI acceleration and pre-integrated 91 open-source AI models, bringing a new level of sophistication to this field. Real-time device analytics can predict failures and prevent downtime while reducing associated costs. Its AI capabilities extend to monitoring conveyor systems, detecting anomalies, and signs of vibration and wear, facilitating predictive maintenance programs to ensure uninterrupted goods flow and minimal worker downtime. Furthermore, its powerful AI processing capabilities can identify worker safety in real-time, ensuring compliance with safety equipment rules by monitoring attire and movements, thus enhancing overall workplace safety.

Radxa ROCK 5A: Real-Time Supply Chain Management at Your Fingertips

In today’s highly interconnected world, the success of supply chains relies on real-time data processing. The Radxa ROCK 5A is powered by the advanced Rockchip RK3588S SoC, featuring an octa-core Arm® DynamIQ™ CPU and Arm Mali™ G610MC4 GPU, ensuring exceptional processing efficiency. Enhanced AI capabilities driven by an onboard 6TOPS NPU improve interactivity and intelligence in digital displays, particularly in computer vision and image processing. HDMI output supports resolutions of up to 8Kp60 for seamless real-time inventory updates, while predictive analytics are displayed via dual micro HDMI ports. With a 40-pin GPIO interface and multifunctional USB Type C™ and HDMI connections, the ROCK 5A can seamlessly integrate with sensors, cameras, and displays, providing a unified data analytics platform. It supports various operating systems, including Android 12 and Debian/Ubuntu Linux, offering flexibility for customized digital display solutions that drive advancements in supply chain technology.

OKdo Leading Supply Chain Innovation

OKdo’s Edge AI solutions are actively reshaping traditional practices. The Tiny Code Reader, ROCK 5 AIO, and ROCK 5A represent practical applications of AI, transforming inventory control, predictive maintenance, and ensuring worker safety. Design engineers in manufacturing find valuable allies in these advanced technologies, paving the way for enhanced efficiency, steadfast reliability, and improved safety protocols in the supply chain.

Human-Machine Collaboration and Worker Safety and Training

With the rise of automation in manufacturing, Human-Machine Collaboration (HRC) has become a critical aspect of optimizing operational efficiency. Edge AI is an effective factor in achieving successful HRC. As part of Edge AI systems, deploying AI models on edge devices enables real-time feedback between robots and humans, facilitating seamless collaboration between human workers and robots.

Real-time feedback can improve synchronization, enhance safety measures, and enable efficient task allocation. Overall, this Edge-AI-driven collaborative approach allows manufacturers to leverage the strengths of both humans and machines, thereby increasing productivity and improving operational outcomes.

An engineer using KUKA robots in a smart manufacturing environment (Image Source: Zenoot)

In terms of safety, Edge AI can utilize AI models to detect hazardous situations in real-time, significantly enhancing worker safety. By continuously monitoring data from sensors and devices, Edge AI systems can timely identify potential risks, such as machine malfunctions, abnormal movements, or hazardous conditions. Alerts and notifications can be sent to workers or supervisors, enabling them to take immediate action and prevent accidents. In this regard, the ability of Edge AI to analyze data on edge devices ensures minimal latency when detecting safety-related issues, thus guaranteeing worker safety in real-time. Another positive aspect is that this enhanced safety brings increased privacy, as there is no need to transmit workers’ personal data off-site (e.g., to cloud data centers).

Furthermore, adapting to technological advancements and improving workforce efficiency hinges on enhancing workers’ skills. Edge AI can effectively contribute by providing real-time feedback during training tasks. By integrating augmented reality (AR) and virtual reality (VR) applications, Edge AI can offer interactive training experiences, allowing workers to learn and practice in a simulated environment. Additionally, real-time feedback based on AI models can help workers understand their performance, identify areas for improvement, and adjust their actions accordingly. This iterative training approach can enhance workers’ skills and knowledge, ultimately boosting productivity.

Overall, the emergence and rise of Edge AI in manufacturing provide a wide range of use cases that can significantly improve operational efficiency, product quality, supply chain management, worker safety, and workforce skills. These improvements largely stem from Edge AI’s ability to process and analyze data at the edge, enabling real-time performance, low latency, and enhanced security.