Mobile Navigation

Data in motion

Putting industrial data to work helps companies make informed decisions

Harnessing device data via integrated asset management, especially for high-performance robots, helps manufacturers improve product quality for optimal operation efficiency.

Manufacturers are successfully moving past the technology tipping point into digital transformation. The adoption of advanced machines and their intelligent peripherals, along with production monitoring, is accelerating this fundamental shift, creating completely integrated smart factories.

Robust robots and their subsequent technologies are being paired with smart sensors and intuitive software to provide a wealth of equipment and production information so companies can make informed decisions efficiently regarding preventive and predictive maintenance, quality assurance, demand forecasting, inventory management, price optimization, warranty analysis, machine learning and more.

Intelligent technology

The industry continues to progress from brand-specific data collection of a single device on a local area network (LAN) to the use of inclusive cloud-monitoring systems that support an array of heterogeneous equipment, including third-party applications. This is due much in part to edge server solutions that reside between the LAN and external internet resources, such as cloud data warehousing and processing. Easily adaptable, these modern-day platforms implement a common connectivity framework or communication protocol that uses intelligent technology for networking devices.

Widely used in manufacturing environments is the feature-rich Open Platform Communication Unified Architecture (OPC-UA) interface. The successor to the OPC (Object Linking and Embedding for Process Control) standard, the unified architecture of the Ethernet-based OPC-UA provides an intuitive method for device interoperability that is independent of proprietary application programming interfaces.

Robots have the ability to capture data related to speed per axis in addition to the amount of time a robot is moving versus idle time.

Quite effective, OPC-UA options, like the Yaskawa Cockpit, facilitate an integrated, intelligent and innovative approach to data harvesting that offers a one-to-one topology where clients (aka, devices) request data and servers timely respond with that data.

Each client can interact with multiple servers and vice versa. At times, a single device may even function as both a client and a server. When needed, additional functionality is available via the OPC-UA Publish/Subscribe (PubSub) AddOn. Either way, this method enables manufacturers to see what is happening at any point on the value creation chain to gain actionable insights for customizing operations that can better fulfill company initiatives.

Companies looking to implement a protocol for low-resource devices in low-bandwidth networks with high latency often turn to MQTT (Message Queue Telemetry Transport). Ideal for embedded computers, low-power sensors, microcontrollers or mobile devices, this lightweight message protocol works via a PubSub model that uses an open-source message broker (aka, a server) such as Eclipse Mosquitto. Essentially, this option enables a server to receive data then distribute it back to subscribers who request data on that particular topic.

Highly debated in the IT community as to which one is the best, each of these, as well as other common communication protocols, can achieve effective data management.

Putting data to use

Whether combined to ensure surface quality, weld integrity or part geometry of an in-house part, integrating various technologies can provide a wealth of data for making informed production decisions.

Implemented well, nearly every company can benefit from integrated asset management – as the digital data accumulated is highly advantageous for monitoring key performance indicators. Along with this, decision makers are finding innovative ways to use machine data for other purposes. For the manufacturing sector, robotic workcells are a prime example of how cumulative data can be used to optimize various applications or aspects of a production process.

For example, welding robots can capture data related to speed per axis, as well as the time a robot is moving versus the time it is idle. Robots can even help provide a graphical representation of voltages applied to axes over time. A robot controller is also a knowledge goldmine for users looking to gain understanding on aspects such as cycle time per job or part, number of cycles completed, arc starts or weld counts, arc-on time, number of contact tip changes, hood filter hours remaining, and the average voltage and current of a welding power supply.

Similarly, power supplies and wire feeders can capture arc starts or weld counts, arc-on time, wire feed speed, amount of wire consumed, and the average current and voltage. Also trackable are workcell cycle starts or stops, along with the number of cell entries. More complex workcells may provide advanced sensor data for environment variabilities like temperature, pressure and humidity, as well as information for mechanical factors pertaining to vibration and acceleration.

Independently, all of this data may or may not be helpful. However, given the proper analysis, it can help maintain overall equipment effectiveness for peak production throughput, while mitigating errors or preventing downtime. Integrated with an array of devices, robotic systems can easily be tracked to provide a treasure trove of insight.

Weld traceability

To better enable serializing and storing of data and source information of any given part, robot suppliers help manufacturers by designing weld interfaces that support arc monitoring capability that power source manufacturers may have integrated. As a result, local and remote data collection is being used to monitor changes in the weld process. Accessible from the robot teach pendant, these interfaces often have documentation and data analysis systems that can connect multiple power sources, enabling precise evaluation of a variety of parameters.

Hundreds of times a second, a power source measures its output voltage, wire feed speed, weld time and specific parameters (when applicable), comparing the data collected to predefined limits for the weld in progress. If the power source identifies any parameter outside of acceptable limits, that weld is marked as “suspect,” and the information related to the weld is transferred to the robot and robot operator for proper handling.

Examples from welding equipment suppliers include Miller Electric’s Insight Centerpoint arc data monitoring software that is available via its smart and powerful Auto-Continuum power source. Lincoln Electric offers Production Monitoring, along with a unique WeldScore feature that ranks the weld’s overall quality based on the previously mentioned criteria. Fronius’ WeldCube connects multiple TPSi welders from one device, and can generate daily production and quality reports for equipment status and health.

Similarly, data can also come from external devices such as laser profile sensors for weld quality inspection, or cameras for reading serial or QR codes. All data is then compared against a predetermined threshold to decide what is an acceptable or flawed part.

Part testing and inspection

Data can come from external devices like laser profile sensors for weld quality inspection.

Advances in artificial intelligence (AI), simulation and modular hardware continue to positively impact robotic applications. Part testing and inspection are no exception. This approach provides manufacturers an objective analysis of a process or a procedure versus fixed criteria to determine if a part process can be verified. Companies can test good parts against bad parts, or process success can be tested versus process failure.

More versatile than costly and inflexible coordinate measuring machines, robots equipped with intelligent peripherals can provide a wealth of data for making informed production decisions. Used in a range of industries including automotive, aerospace, electronics and plastics – especially when safety-critical parts are involved – robotic part testing and inspection is highly valuable for ensuring surface quality, weld integrity and part geometry of an in-house part. It can also be utilized to check the caliber of parts being brought in from another part of the supply chain.

While various technologies can be implemented to achieve this, newer methods incorporating sound analytics continue to gain traction. With all devices synchronized via an edge server solution, a welding workcell – incorporated with a high-performance robot, welding power supply, 2-D machine vision, sound analytics platform and robust yet user-friendly weld inspection system – can expertly detect part anomalies via trained acoustic insights.

Done in a non-destructive manner that proactively remedies quality problems for greater production yield, this process is ideal for determining product defects early in the manufacturing process. Furthermore, this mix of innovative technologies and cognitive computing delivers significant costs savings by reducing scrap, rework and warranty costs.

AI learning

Technology advancements continue to enhance robot dexterity, creating more flexible, skillful and easier-to-use robotic systems. Combining intelligent machine hardware with AI advancements, manufacturers are addressing the simulation-to-reality (Sim2Real) gap. Learning data that highly resembles a real environment can now be generated via the use of a simulator, enabling highly efficient operations including quick implementation of a robotic system.

For the expedient handling of soft or irregular objects located haphazardly or stacked in a bin with varying orientation, AI-driven part and stacking data generated from the simulation process effectively teaches a robot which path and points it can realistically and stably accommodate. Companies, such as Ambi Robotics and Mujin develop advanced solutions to meet high handling demand, while empowering humans to work smarter.

Ambi’s AI-powered Ambisort solution uses the intelligent AmbiOS operating system to train a robotic system to be 10,000 times faster than alternative solutions for increased adaptability. Similarly, Mujin offers an all-purpose intelligent robot control system for more capability, efficiency and reliability for improved productivity. From upstream processes to final loading, the company’s patented 3-D perception system with AI technology accelerates bin picking, insertion, piece picking, palletizing and depalletizing tasks.

For welding, Path Robotics is a leader in intelligent path planning. Featuring a proprietary version of AI software utilizing camera imaging, Path’s turnkey solutions use patented technology to determine part shape and weld seam locations. Also giving a QA analysis of weld performance, this concept provides users access to the equivalent of many skilled welders. Typically using direct robot interfacing, these systems offer preselected weld settings that allow one-off parts for high-mix applications.

The daunting task of figuring out how to connect a multitude of different devices is becoming a thing of past, and standardization efforts of network protocols and communication abilities between devices within industrial environments continue to improve. Manufacturers that successfully implement robotic automation and harness the data available should be able to optimize product quality and smooth production woes, building a strong competitive edge.

Yaskawa America Inc.