Collaborative Intelligence, Seamless Human-Robot Integration

The Development of Collaborative Robots and Taiwan's Competitiveness

A Collaborative Robot (Co-robot or Cobot) is defined as "a robot designed to operate in a shared workspace and interact directly and physically with humans in close proximity." Unlike traditional industrial robots, cobots prioritize safety and flexibility, allowing them to work alongside humans while boosting production efficiency. With the continuous evolution of Artificial Intelligence (AI) and smart machinery technology, the application scope of collaborative robots is gradually expanding from manufacturing to broader areas, including industrial production, logistics, medical care, and retail services, demonstrating strong development momentum.

According to a market research report published by the International Federation of Robotics (IFR) on December 4, 2024: "541,302 industrial robots were installed globally in 2023, with collaborative robots accounting for 10.5%." This data shows that cobots hold a significant position in the wave of automation and productivity enhancement and continue to receive high market attention. In the future, driven by AI, sensing technology, and smart control, collaborative robots will not only be key for enterprises to boost competitiveness but also effectively mitigate the challenges of labor shortages. Perhaps, when there aren't enough hands, robots can indeed step in to fill the gap.

Classification of Collaborative Robots

Based on the standards set by the International Organization for Standardization (ISO) 10218, collaborative robots can be clearly divided into the following four types:

1. Hand Guiding This type of cobot allows an operator to physically guide the robot to complete a specific task, simplifying the programming process and reducing downtime. Through intuitive guiding, the operator can quickly adjust the robot's movements without changing software or using complex programming tools.

2. Safety Monitored Stop This type of cobot uses sensors to detect the presence of a human. When a person enters a predetermined area, the robot automatically pauses operation. Once the area is clear, the operator only needs to press a button to restart the robot. This is suitable for industrial automation tasks with minimal human-robot interaction requirements.

3. Power and Force Limiting (PFL) Safety is a priority in the design of these cobots. They typically feature rounded edges to avoid sharp points and have built-in force limiting mechanisms that automatically stop upon contact with a human. They do not have exposed motors or pinch points and are generally smaller, slower, and more flexible, making them suitable for close human-robot interaction where extra safety measures are needed.

4. Speed and Separation Monitoring Equipped with advanced sensors and machine vision systems, these cobots maintain a safe separation distance from humans. Operation is divided into two states based on area segmentation: when a human approaches, the robot automatically slows down; when a human enters the stop zone, the robot halts completely and automatically resumes work once the person leaves.

Safety Standards and Regulations

After understanding the classification of cobots, it is evident that their application can be further categorized based on the depth of human-robot interaction, ease of operation, and whether the operator requires professional training. However, as robots increasingly enter the workplace, enterprises must not only consider production efficiency but also ensure compliance with labor laws and workplace safety standards when introducing such equipment.

Many people are familiar with the Labor Standards Act protecting human workers, but did you know that Taiwan's Ministry of Labor (MOL) has also formulated relevant regulations for the entry of robots into the workplace? These regulations address the trend and potential risks of human-robot collaborative work.

The MOL amended and promulgated the "Standard for Prevention of Hazards from Industrial Robots" on February 14, 2018, explicitly requiring employers using collaborative robots to comply with the National Standard CNS 14490 series, the International Standard ISO 10218 series, or equivalent standards. Employers must conduct a safety assessment based on relevant data and prepare a safety assessment report for inspection.

To help enterprises implement this regulation, the MOL further released the "Key Points for Safety Assessment of Collaborative Robot Operations" on March 27 of the same year, detailing the assessment items to ensure the safe use of collaborative robots. These measures are not just to reduce the potential risks posed by robots to personnel; they are also a response to the trend of human-robot coexistence in the workplace, ensuring that enterprises and workers can fully leverage the potential of cobots in a safe environment.

• National Standard CNS 14490 Series: This is Taiwan's safety standard specifically for collaborative robots. Employers must follow these standards and conduct corresponding safety assessments when using cobots.

• International Standard ISO 10218 Series: This is a set of international safety standards for industrial robots. Collaborative robots must also comply with these standards to ensure safety during interaction with humans.

Integration with Discriminative AI and Application Benefits

Collaborative robots are designed to be lighter than traditional industrial robots, primarily to enhance flexibility and allow them to quickly adapt to different production environments. Cobots are also suitable for many applications that do not require direct human-robot interaction and play a key role in the field of smart manufacturing. As cobots become more common in manufacturing, AI technology is also being integrated into the sector, particularly in applications like computer vision, predictive maintenance, and automated decision-making, driving smart manufacturing forward.

The combination of Discriminative AI and collaborative robots is particularly effective. Discriminative AI is trained using massive amounts of data to enhance the robot's analysis and real-time decision-making capabilities, allowing it to more accurately identify objects, optimize operating trajectories, and perform anomaly detection and quality control, thereby ensuring the stability and precision of the production process. Currently, AI technology is applied in scenarios such as machine vision inspection (e.g., defect screening) and smart logistics (e.g., automated handling), and is gradually expanding into more complex manufacturing segments.

Next, we will explore the application benefits of collaborative robots combined with Discriminative AI:

• Intelligent Operation and Collaboration The combination of collaborative robots and Discriminative AI allows for a more precise understanding and adaptation to the working environment, improving production efficiency and flexibility. For example, in electronic assembly and automotive manufacturing, AI can help robots perform real-time anomaly detection, improving product quality and production stability. Furthermore, human-robot collaboration optimized by AI allows operators to focus on higher value-added tasks, such as product design and process optimization.

• Data Analysis and Real-Time Decision-Making AI can analyze data collected by robots to provide real-time decisions, helping enterprises quickly respond to production needs and market changes. For instance, some robots are already equipped with equipment health diagnostics and predictive maintenance technology. Based on detected data, they can predict mechanical failures, reducing downtime, improving operational efficiency, and lowering maintenance costs.

• Versatility and Flexible Deployment Collaborative robots can perform multiple tasks, including material handling, assembly, and quality inspection, and can be quickly reprogrammed as needed to adapt to different working scenarios. Moreover, through Plug-and-Play technology, cobots can be flexibly deployed in various production environments, making them particularly suitable for high-mix, low-volume or highly variable production models, which is more appealing to small and medium-sized enterprises (SMEs) and non-professional engineers.

• Safety and Adaptability Collaborative robots usually have built-in collision sensing and safety mechanisms, allowing them to operate in a shared human-robot space. However, in certain application scenarios, they still need to comply with international safety standards like ISO 10218 and ISO/TS 15066, and may require additional virtual fences or sensors to ensure operational safety.

The integration of collaborative robots and Discriminative AI is a major trend in the future development of the manufacturing industry, promising to further improve production efficiency, quality control, and human-robot collaboration, realizing a comprehensive upgrade to smart manufacturing. Several related technological achievements have already emerged:

1. "High-Fidelity Human-Like Dual-Arm Collaborative Robot" - Boosting Taiwan's Hand Tool Industry Upgrade This technology uses AI to automatically generate assembly procedures, enabling the robot to precisely mimic the movements of a technician, flexibly operating various hand tools with two arms. The robot can flexibly assemble over 120 types of hand tools and effectively reduce jig costs by 50%. This not only significantly boosts production efficiency but also adapts quickly to the domestic demand for flexible smart manufacturing through rapid customization, making production lines more elastic and competitive. This is leading a manufacturing revolution in Taiwan's hand tool industry! It has successfully partnered with the domestic hand tool "invisible champion," Yingfa Enterprise, and the technology has been successfully introduced into their production lines.

2. "HolonOS" - Driving Industry 5.0, AI-Generated Optimal Work Paths, Creating New Standards for Smart Manufacturing Facing the challenges of high-mix, low-volume manufacturing, coupled with labor shortages and increasingly complex equipment operation, ITRI launched the smart manufacturing solution "HolonOS." It uses AI to automatically generate the optimal work path and compensates for variations in real-time, enhancing production flexibility and efficiency. Utilizing an intuitive PC-based operating interface, it supports major robot brands both domestically and internationally. HolonOS not only solves the difficulties of cross-factory and cross-equipment integration but also centralizes the management of process data, ensuring production stability and precision. Through AI smart learning, HolonOS can significantly shorten the training time for new personnel by 50% and retain 90% of a company's proprietary process technology, maximizing the benefits of human-robot collaboration and making production lines more competitive. HolonOS has since been spun off by ITRI into a startup company, Holon (赫侖), and successfully transferred its technology to industries such as water hardware, kitchenware, high-end cutlery, bicycles, hand tools, and auto/motorcycle parts, further expanding into the semiconductor and aerospace sectors.

3. "RGB-D AI Robot" - Autonomous Learning, Creating New Standards for Smart Picking Combining AI autonomous learning technology, the RGB-D AI Robot is like giving the robotic arm a "brain." It can automatically label tens of thousands of training data points and quickly learn the optimal picking strategy within 12 hours. As the AI capability improves, the robotic arm can instantly identify new objects and precisely determine the gripping or suction point, greatly enhancing operational flexibility and stability. By having AI automatically generate high-fidelity, high-quality, and large amounts of labeled training data, the RGB-D AI Robot can quickly adapt to and perform precise picking when faced with new object styles without the need for re-teaching. This significantly reduces human intervention and training costs, bringing a new breakthrough to smart logistics and automated production.

4. "RobotSmith" - Aiding Taiwan's Industry to Advance Towards Digitalization and Intelligence This is the nation's first virtual-real integrated robotic production line interaction assistance system. Through proprietary software integrating sensing technology, robot technology, and industry knowledge, it provides a comprehensive, high-precision, and high-efficiency solution for metal surface finishing. The system can automatically perform trajectory planning and optimize path parameters, ensuring processing quality and production efficiency. RobotSmith boasts three core advantages: 1. Automatically generating robot processing programs, shortening the teaching time by 95%; 2. No programming required, deploying the production line manufacturing process in one second using voice control; 3. Zero-latency presentation, predicting sandpaper wear with over 95% accuracy. Furthermore, RobotSmith is compatible with over ten international robot brands, quickly achieving virtual-real integration and collaborating with experienced technicians to truly implement smart manufacturing!

I recall that when robots entered the manufacturing industry, negative public sentiment surrounding their large-scale adoption grew, fearing widespread job losses. However, just as humans are not perfect, neither are robots. It is well known that operating a robot is not as simple as turning on a computer and running a program. Robots need to have the ability to observe their surroundings and listen, which requires a vast number of sensors and years of development and validation to achieve complex actions such as a robot picking up an egg. Although many humans are born with the ability to pick up an egg without breaking it, achieving this capability in robot R&D requires enormous effort and study. First, the robot must be able to recognize that the object in front of it is an egg and ensure that the force applied by its fingers does not break the egg. Consider this: if a truly automated "unmanned factory" were realized, completely eliminating the need for humans, would the associated benefits and cost recovery truly be worthwhile? This is undoubtedly a question worth pondering deeply.

Therefore, viewing the bigger picture through this small example, achieving highly efficient collaboration in production line layout requires the simultaneous presence of both humans and robots—neither can be absent. Humans can compensate for the robot's deficiencies in vision, touch, and hearing, while the robot can provide humans with support in terms of physical strength, endurance, and meeting the demands of high-intensity work. By retaining the respective strengths that are difficult for the other to achieve quickly, and by emphasizing the spirit of mutual assistance among people, why not extend this concept to human-robot collaboration? This approach would not only allow domestic SMEs to balance the cost recovery of robot investment but also enhance the application benefits of the robots. Leveraging the best of both sides, is this not a more effective choice for enterprises?

As the saying goes, "Two heads are better than one." Future collaborative robots will achieve more natural interaction with humans through more advanced sensors and AI technology. This will enable robots to better understand human intentions and behavior and collaborate seamlessly with humans in complex working environments, truly achieving "human-to-human integration" between humans and robots. Collaborative robots represent not just a technological upgrade but a critical key to the transformation of a company's production model. To ensure the maximization of introduction benefits, enterprises should adopt a data-driven decision-making approach. Specifically, they should clearly define goals and requirements, conduct precise assessments (such as production line suitability and employee training), and implement small-scale pilots and iterative optimization. Furthermore, establishing long-term maintenance, operation, and upgrade strategies is crucial. Through these measures and strategies, enterprises can reduce the risk of adoption and increase their return on investment, allowing them to gain a head start in the future competitive landscape.