November 14, 2016 by Treena Hein
Nov. 14, 2016 – Analytics are found in most facets of our society today, and manufacturing is no exception. Analytics — colossal and powerful software programs able to automate or semi-automate tasks and decisions — are becoming more prevalent in plants across Canada and around the world, and for good reason.
Simply put, the use of analytics in manufacturing provides improved machine reliability, product quality, process flexibility and assembly speed by enabling machines to intelligently coordinate and collaborate in a flexible manner in real time. One new analytics system expected to take manufacturing automation to a new level is the Fanuc Intelligent Edge Link and Drive (FIELD), created by Japan-based global robotics specialist Fanuc. Partners in its development include Cisco (digitization), Rockwell Automation (industrial automation) and Preferred Networks (artificial intelligence or AI). Currently being introduced to a circle of suppliers and developers, FIELD is expected to be fully launched by the end of 2016.
The FIELD platform combines AI and huge computational power to analyze data generated by robots and peripheral devices used in tasks such as bin picking, anomaly detection and failure prediction. In addition, it’s also an open platform, so application developers, sensor and peripheral device makers, system integrators and others can use it as a launch pad to build solutions to help improve their equipment efficiency, manufacturing output and product quality. Some applications will be offered in a continuously growing portfolio to other robotics suppliers for download to their own robots, much like one downloads an app on a smartphone, says Fanuc.
While it is one of several existing platforms for integrating the automation of industrial systems, some experts believe FIELD could become the standard for fully automating factories, and that it will allow much faster integration of robots and digital technologies compared to the existing pace of adoption. For its part, Fanuc says FIELD is nothing less than a system that will result “in sophisticated manufacturing practices not before possible.”
Nuts and bolts
Fanuc executive vice president Dr. K. Inaba had the original idea for the FIELD system. “He envisioned the need for a common open platform for the manufacturing sector for automation equipment such as CNC machines, robots, PLCs and sensors to easily connect and intelligently utilize the vast data that is provided by these machines,” explains Jason Tsai, vice-president of product development, Fanuc America. “The FIELD system will provide a platform that will allow applications such as ‘zero down time’ or capabilities like deep learning or machine learning to be applied in the ‘edge level’ within the factory site.” ‘Edge level’ or ‘edge heavy’ processing refers processing huge amounts of data physically at the manufacturing site, so that the costs involved with transmitting large volumes of data to the Cloud and back are minimized. However, a secure connection to the Cloud is kept in place for applications that are suited to that.
Tsai says FIELD will take manufacturing automation to the next level by allowing devices to intelligently coordinate and collaborate in a flexible manner to achieve sophisticated manufacturing practices. Basically, it all comes down to the way FIELD connects parts of a factory to its central analytics ‘brain,’ so that incoming data from many sources can be analyzed on an ongoing basis by the analytics program. The analytics can therefore produce an updated ‘big picture’ on which improved decisions can be made.
“Once the FIELD software (both open source and proprietary licensable products from Cisco and Preferred Network) is loaded and configured,” Tsai notes, “the machine data from all the automation equipment — including CNC machines and robots — can then be accessible via the computing devices using a suite standard industrial protocols.”
Multiple types of machines (including computing devices) are attached to the different levels of the FIELD network, some added to the edge level and some added to production cell levels so message distribution of machine data is managed in an efficient and effective way. “This allows a third-party company to develop the functionality which requires communicating the real-time machine data among all FIELD computers,” says Tsai.
With all the communications involved in FIELD, potentially among hundreds of devices, a key feature of the system is the way it strives to maintain secure communication. Working with Cisco, the FIELD system promises to provide network security that is essential to the manufacturing production operation, where the performance and integrity of the network must be maintained, Tsai explains.
In terms of its impact on optimizing production at the time FIELD is installed and into the future as well, Tsai notes that the data on the FIELD system from various machines is used to do a variety of tasks, from setting up collaborations among equipment, optimizing individual asset performance, and to improve both tractability and quality control. In addition, having the system architecture in place on the manufacturing floor significantly increases the potential to incorporate new automation capabilities.
Tsai adds that FIELD is unparalleled in enabling “zero down time” through its ability to monitor vast amounts of data provided by the equipment to predict and detect failures. And he reminds us that “in addition to applications that Fanuc will provide for their own machines such as CNCs and robots, since FIELD system is an open platform, it will allow companies to also make their own similar kinds of applications to monitor their own machines.” Fanuc believes many firms will use FIELD to develop new and creative digital functions to support intelligent capabilities using external sensors and other devices. Tsai says FIELD “absolutely” provides manufacturers with ways to write code that they couldn’t otherwise write.
Much of FIELD’s power lies in its deep learning algorithms, developed by Fanuc in partnership with Preferred Networks. An example of deep learning implementation is robotic bin-picking applications — selecting a single part at the right orientation from a pile of similar parts laying randomly in a bin. “Bin picking has been considered a challenging application for robots, requiring skilled programmers,” Tsai explains. “By utilizing Preferred Network’s deep learning technology, and having the FIELD system’s edge-heavy processing provide the computing power, Fanuc has been able to improve the bin pick success rate significantly with minimal teaching to the robot. The robots would learn by itself which part in the bin is positioned most optimal for picking through multiple trials.” The deep learning includes “distributed learning capabilities,” which means that robots on the FIELD network can share their knowledge, which significantly reduces bin picking learning time and provides other advantages.
Rowan Trollope, senior vice president of Internet of Things (IoT) and Applications at Cisco, is unequivocal in his belief that FIELD is a true manufacturing game-changer.
“This collaboration represents a historic shift in the industry, with IoT, industrial automation and machine learning coming together to make the factory of the future a reality,” he stated in a recent Fanuc press release. “It’s been talked about for years, but now it is really happening.”
Treena Hein is an award-winning Ontario freelance science and tech writer.