Robots Become More Useful In Factories

Most people associate factory automation with large robotic machines, such as those that weld automobile chassis on assembly lines. But as prices drop and technology improves, robots are being deployed for smaller and more varied tasks, and they are getting better at all of them.

Inside of factories, robots can significantly improve output, consistency, and reliability. They can work around the clock, and for certain applications, they can do the job faster than humans, with higher precision, and with fewer errors. Some robots can lift up to 500 kilograms (1,100 pounds), reducing both the amount of manual labor needed and downtime due to injuries.

In the past, robots typically required a large up-front investment. That’s no longer the case. As with the semiconductors that control them, economies of scale have brought the startup costs within the budgets of even small companies. A turnkey solution service for machine task automation, including hardware, software, and support, can be implemented for as little as $25,000 per year.

This makes the goal of smart manufacturing, also known as Industry 4.0, significantly easier to achieve. And it can improved further with automation tools and artificial intelligence.

Manufacturing productivity improvements also can reach well beyond just automated assembly. A factory consists of many departments, including management and process planning, accounting, engineering, assembly, material and warehouse management, receiving and shipping. Automation is an important element in the production formula, but it also can be integrated across all those departments.

There is even a formula for this:

Factory productivity = Process Planning Including Production Flow + Manual Labor + Full-automation of Subtasks + Semi-automation of Subtasks.

“Robotics offers a lot of benefits in automation,” said Chetan Khona, director of Industrial, Vision, Healthcare & Sciences at AMD. “Given the current worker shortage worldwide, these machines can help augment production to meet demand throughout the supply chain. They can incorporate machine vision, artificial intelligence, and a variety of sensors to work in factories, as well as harsh and remote areas, where they operate very fast and can do things more safely than with humans.”

This isn’t a trivial task, however. “Designing such machines, can be very involved,” Khona said. “It includes software programming, hardware, and system integration. On many occasions, using adaptive SoCs and SOMs will cut down design time and increase reliability. Additionally, they deliver high performance with integration of all features and functions in a single platform.”

Gartner defines robotic process automation (RPA) as a “market for licensed software platforms used for building scripts to integrate any application via a user interface and a control dashboard or orchestrator. RPA platforms automate repetitive, rule-based, predictable tasks.”

These bots can be programmed with a low-code/no-code graphic user interface (GUI). Gartner envisions this segment to be one of the fastest growing markets. Key players include IBM, SAP, and Microsoft.

“The main focus of manufacturing is to increase productivity measured in throughput over a time period, with minimum downtime,” said Sathishkumar Balasubramanian, head of product management and marketing for IC verification at Siemens EDA. “But assembly line manufacturing line is a dynamic environment, and automation is only part of the solution. On the outside, it seems to be important to have constant flow. However, variability in manufacturing flow is inevitable, and how the manufacturing process adapts to variation is highly critical to keep the downtime to a minimum. For example, in bottling manufacturing, how the work moves from station 1 to station 4, and a change in bottle orientation, can be addressed by an adaptive production line to meet peak demand with minimum disruption. That is very important. The ability to sense the status of manufacturing line at the edge is key to robotic manufacturing process.”

Some level of autonomy can be added into the various steps to improve productivity, as well. “Each station in the manufacturing line should be capable of making decisions, and communication across entire manufacturing line is critical to further improve efficiency as part of Industry 4.0 evolution,” Balasubramanian said. “In the end, designing the overall factory floor with an intelligent and efficient manufacturing flow to achieve the best ROI is the ultimate goal. To achieve this, the ability to replicate the entire manufacturing process, including the individual stations as a digital twin, is key.”

How do robots function?
The key components of an industrial robot include controllers — these can be hardware, software, and they can include various processing elements such as GPUs MCUs, and/or AI coprocessors — as well as sensors (vision, motion, and distance sensing), motors, robot arms, and end effectors. Sometimes, an end effector is referred to as “end of arm tooling.” A welding tool, a gripper, a dispenser, or a soldering gun are examples of end effectors. Robots can be considered as automation tools much like other manufacturing tools to facilitate the manufacturing process.

The International Organization for Standardization (ISO) published a document ISO 9283:1998 – Manipulating industrial robots — Performance criteria and related test methods to help companies evaluate robot performance. Some of the performance criteria include pose accuracy and pose repeatability, distance accuracy and distance repeatability, position overshoot, path accuracy and path repeatability, and path velocity characteristics.

Fig 1: Working alongside robots. Source: Siemens

Well-suited for fast and repetitive tasks, industrial robots can be programmed to do different manufacturing jobs. Some examples are processing, pick and place — including sorting — warehouse fetch and delivery, packaging, and assembly.

“Designing such systems involves something we call specialized processing,” said Suraj Gajendra, senior director of technology strategy for Automotive and IoT at Arm. “The robotic arm has to be designed to meet a great level of determinism in its operation from both a real-time perspective and for functional safety. Each motion algorithm the robot follows must be executed within its allocated time-window without fail, and in any case where it malfunctions, the robot should automatically be brought into a ‘safe-state’ where human intervention is possible.”

This works like any centralized compute architecture, but with intelligence added into the sensors. “Getting a high-precision robotic arm functional involves multiple sensors feeding all critical data to a centralized processing unit,” Gajendra said. “They can be motion sensors, pressure and positioning sensors, camera and visual sensors, etc. The central processor makes real-time decisions based on a combination of pre-programmed inference algorithms and the sensor data. Most of the time, these decisions are taken locally by the processor running the robot.”

Types of robots
As of today, there are three main types of robots — Cartesian, selective-compliance-articulated robot arms (SCARA) and six-axis robots.

Cartesian robots, which are the most precise, usually are mounted on the floor and use linear actuators to move in the x, y and z directions. SCARA robots are similar to Cartesian robots, but with an additional theta axis, enabling rotation within the Z-plane. In contrast, six-axis robots — also called six degrees of freedom — can use their ability to move in six directions, (up, down, forward, backward, and wrist turning of the robot arm), to simulate a human arm.

These robot arms can be small enough for installation at stations alongside human workers, and used to augment tasks performed by humans. In this hybrid model, the collaborative robots are called cobots, and adoption is growing as they demonstrate their usefulness.

There also are a growing number of other robotics types in development or more limited use, including articulated, polar, delta, and cylindrical, all of which enable additional movement or rotation.

Humanoid robots are gaining in popularity, as well. C-3PO and R2-D2, the human-like robots featured in the movie Star Wars, greatly increased awareness of robots. Four years ago, a human-like, social robot named Sophia (two-minute video) demonstrated that an AI-enabled robot could learn and carry on a conversation with human beings.

There has been significant progress since then. Some hospitals have started to experiment with human-sized social robotic nurses to entertain patients. Someday, they may even join the factory work force.

Robotic metrics
The robotic industry has established a set of guidelines known as the LOSTPED parameters. They represent load, orientation, speed, travel, precision, environment, and duty cycle.

1. Load. This is defined as the workload/payload capacity of the robot within a factory. In order to maintain balance, the robot’s weight must exceed the payload, together with any attached tools at the end of the robot arm. For example, if the work load is 150 kilograms, only the largest SCARAs or six-axis robots will be able to do the job. A typical 11 Kg commercial robot arm can only lift a payload of 3 Kg. In addition to the tonnage consideration, for a smaller workload, Cartesian robots can achieve precision within 10 µm, which 100 times better than the other types of robots.
2. Orientation. This relates to how the robots are mounted in the space in which the robot arms will move. Cartesian robots are mounted on pedestals which, while sturdy and able to provide precision placement, take up more space. For smaller work spaces, SCARA and six-axis robots are more flexible.
3. Speed. Robotic arm speed can reach 5 meters per second, with a reaction time of 50 ms.
4. Travel. This refers to the distance the robotic arm can reach. Some can reach up to 20 meters or more, depending upon the application. For assembly workstations in a factory, the travel can be within arm’s length.
5. Precision. Requirements depend upon the application. For pick-and-place in the warehouse, so long as the packages can be picked up safely, the precision requirement is tolerant. On the other hand, certain assembly work requires high precision. Some robotic arms can achieve repeated accuracy of 0.2 mm. When surface-mount technology (SMT) machines are used for flip chip placement with the C4 (controlled collapse chip connection) process, the precision may reach ±25 µm.
6. Environment. Factories have different working environments. Some robotics are most suitable to work in a hazardous or rugged environment, while others are better suited for cleanroom operations. Robots may need IP65 compliance — a rating system on how devices withstand dust and water — to operate in a rugged or dusty environment or to meet the minimum dust and water entry standard.
7. Duty cycle. One cycle of operation, which may be defined as picking up one component and placing it on another assembly, is a duty cycle. Even though robots can work 24/7, nonstop operation will decrease the machine’s life and therefore shorten the mean time between failure (MTBF). It is important to understand robots’ specifications.

Warehouse robots

Fig. 2: Pick and place robot. Source: Righthand Robotics

More companies are deploying warehouse robots to increase productivity. In early 2022, FedEx Express in China installed the DoraSorter AI-powered sorting robot, which can carry 10 kg of packages to 100 destination slots. Around the same time, DHL Supply Chain invested $15 million in multi-purpose mobile robot to automate warehouse operations and improve workflow.

All of this is the result of improvements in AI and increasingly commoditized semiconductor technology. “AI can utilize voice, vision, and sensor analysis to provide solutions to increase productivity and safety in a manufacturing environment,” said Markus Levy, vice president of business development at Kinara.ai. “For example, instead of using human workers to check on inventory in the warehouse, a drone equipped with embedded AI can do the job in less time and more accurately. To design these embedded systems, conserving power is a critical factor. Instead of using a 30W AI processor, which can cost about $500, the design can be optimized with a low-cost $8 to 15 host processor combined with a 600- to 800-MHz AI co-processor to do the system solution, including inferencing. This class of AI coprocessor only consumes 2.5W and has inference latency on the order of 1.8ms to 10ms, depending on the AI model.”

The changing world of robots
Cobots are becoming increasingly popular, exposing more workers to robots and helping them understand their capabilities.

Most robots require programming to perform motions, including linear, circular, and point-to-point. Additionally, speed as well as acceleration and deceleration, need to be considered. By incorporating AI, these cobots can self-learn, and over time they can perform tasks autonomously. Built-in sensors can sense the environment and learn about its space and depth. This class of robots is able to learn and perform multiple tasks and work side by side with humans. Together, the human and cobot team can work much faster than humans or robots working alone, to improve the productivity of the workstation.

“Some manufacturing workstations can be fully automated, but not everything on the factory floor,” noted Anoop Saha, senior manager for strategy and business development at Siemens EDA. “Robots cannot replace humans 100%. They can augment human workers to increase productivity. With 3-D vision and better perception, machines can operate with high precision and reduce errors. But designing and customizing such human–machine solutions requires skills. For example, machines can capture data much faster and improve productivity in the long run with analytics. The data generated need to be used to generate appropriate response, and protected with built-in security. In a robotic factory, robots always need to be connected with secure Industrial 5G, and sensors.”

What’s next
Industry 5.0 is coming. In early 2022, the European Commission (EC) published the Industry 5.0, a transformative vision for Europe report. Its main purpose is to provide “drivers for a transition to a sustainable, human-centric and resilient European industry.”

The EC believes that Industry 4.0 is inadequate. The new vision is to place the well-being of the workers at the center of the production process, with workers using various technologies rather than having technologies replacing workers. This has shed some light on the ongoing debate about whether robots eventually will replace human workers. According to the report, the narrow Industry 4.0 vision focused primarily on productivity. Industry 5.0 would expand the vision to drive technological transformation to achieve people-planet-prosperity.

Industry 5.0 certainly will propel cobot growth in future years, painting a picture of factory automation with increased and measurable productivity.