Since 1993, the IndustryWeek's Technology Leader of the Year feature has served as our platform to celebrate individuals who have successfully leveraged technology to reshape the manufacturing world. Manufacturing greats including Bill Gates, Elon Musk, Harley-Davidson's Jeff Bleustein and GM's Ralph Syzgenda have each taken home the honors.
As the year of the great digital transformation, 2019 saw some of the biggest advancements yet. After all these years of talk about the Internet of Things (IoT) and digital transformation defining the future health and progress for the manufacturing industry, the hype has finally delivered on its promise. Along the way, one company has established itself as the clear leader: Amazon. And, if the future of manufacturing depends on technology and IoT, then Amazon CTO Werner Vogels is the person that will make it possible.
Often referred to as one of the fathers of cloud computing, few others have had as much impact on the future of manufacturing as Vogels. This is especially true as Industry 4.0 has put an emphasis on the importance of digital technologies.
While the journey to revamp manufacturing may have started with cloud computing, it is the blend of data-driven technologies that positions Amazon and Vogels as the leader in this space. Specifically, it is technologies such data analytics, artificial intelligence, machine learning, IoT and edge computing that are ultimately enabling manufacturers to find success within the growing experience economy.
Taking the reigns
As Amazon’s CTO, Vogels is responsible for driving the company's customer-centric technology vision. Vogels joined Amazon in 2004 as the director of systems research after spending time as a distributed systems researcher at Cornell University. He assumed the roles of CTO and vice president the next year, and quickly surfaced as a key force behind Amazon’s approach to cloud computing.
When Amazon Web Services (AWS) first started offering its suite of transformative technologies (2004-2005) outside the company, Vogels’ team was primarily focused on collaborating with companies that needed to quickly achieve internet scale. In those days, they were mostly younger businesses or truly progressive enterprises that already had a clear digital future.
As such, Vogels acknowledges that manufacturing companies were not necessarily an immediate target – and for good reason. At that time, manufacturers were overwhelmingly relying on decades-old equipment that was often singularly dedicated and lacking the ability to generate useable data. However, it did not take long for a digital evolution to transpire within the manufacturing sector.
“It happened that many manufacturers were looking to completely revamp their infrastructure,” says Vogels. “They were interested in using cloud technology to make use of analytics as a key part of being smart manufacturers. That understandably meant finding ways to turn these singular devices into data generators, seamlessly moving the data to cloud, exploring machine learning and then capitalizing on the ability to push insights back into the plant environment.”
Collaborative development…on the edge
Vogels has since worked closely with numerous manufacturers, taking the time to understand their current positions as well as their goals in embracing technology in order to capitalize on the digital realm. “We have always had strong relationships with our customers and together we have developed solutions,” he says. “We do not go off into a high tower, build something and then give it to manufacturing customers. We learn about what they want their technology to look in five to 10 years, and make sure that we build tools that are ready for the manufacturer of the future. This introduced a whole new realm of functionality – especially around device management, security and edge computing.”
The cloud evolution has really moved beyond the phase of companies moving everything to the cloud with people abandoning all local processing. “For quite a few internet and consumer driven companies that is the case. That is their reality,” says Vogels. Yet, for manufacturing, the edge computing component is becoming extremely important. “Manufacturers need to be able to operate in a disconnected mode, but still need to benefit from advanced processing capabilities,” he says. “This creates a very interesting challenge to collaboratively build technologies for devices that will spend a spectrum of time not just in the cloud, but on premise as well.”
Vogels points to Amazon’s IoT Greengrass offering as a direct result of collaboration with manufacturers. Greengrass seamlessly extends AWS to edge devices so they can act locally on the data they generate, while still using the cloud for management, analytics, and durable storage. Whether or not there is an internet connection, devices can run AWS Lambda functions, execute predictions based on machine learning models, keep device data in sync, and communicate with other devices securely.
Manufacturers can get rich insights at a lower cost by programming devices to filter data locally and only transmit the data needed for applications to the cloud. Understandably, the goal is to reduce the amount of raw data the manufacturer sends to the cloud, which ultimately minimizes the cost and increases the quality of the data transmitted. AWS is going a step further by integrating machine learning into edge devices. ML uses algorithms that learn from existing data, a process called training, to make decisions about new data, a process called inference. Because inference requires significantly less computing power than training and optimizing ML models, it occurs in real time when new data is available.
According to Vogels, getting inference results with low latency is important for making sure that IoT applications respond quickly to local events. Such local inference could allow a robot to make autonomous decisions in near-real time, even without a connection to the cloud.
Recognizing growing digital potential
Although AI is very a broad concept, the machine learning component and the ability to make use of the very large data sets continues to show potential, especially when considering the phenomenal amounts of data US manufacturers create daily. “That is only going to grow tremendously, both in volume and importance,” says Vogels.
Woodside Energy serves as a prime example. Initially, Woodside’s Pluto Liquefied Natural Gas facility in Western Australia leveraged 10,000 sensors, primarily capable of alerting key personnel when faults occurred. However, being able to leverage cloud-based machine learning capabilities has proved transformative.
Today, Woodside uses sensor data to build an algorithm that allows the team to predict and prevent foaming in the Acid Gas Removal Unit (AGRU), a critical part of the production process that cannot be monitored directly. Currently, Woodside runs more than 6,000 algorithms on the sensor data from its Pluto plant. These same operations now make use of 200,000 sensors, monitor operations 24/7. By connecting these IoT sensors to the AWS Cloud, Woodside has been able to optimize the production and maintenance. “This is a major shift from reacting to an alarm going off to being able to take data, analyze and immediately predicting potential issues,” says Vogels.
Access to new datapoints is enabling Woodside’s data science team to continuously find new insights by sharing data across the organization. Of course, the Woodside transformation is ongoing including a dedication to explore uses for AI to augment and inform better decision-making.
Data fueled future
According to Vogels, many of the truly smart manufacturers have already passed through the barrier of Industry 4.0. And, some of these manufacturers have significant plans to fully capitalize on the digital realm. For example, in March 2019, AWS announced an ongoing cooperation agreement with Volkswagen (VW). As part of agreement AWS is building an “industrial cloud” for VW, which will connect all of VW’s 122 global factories, enabling the automaker to seamlessly collect all of its data in real-time and put it into the cloud.
As VW’s industrial cloud solidifies, the plan is to use the AWS IoT, machine learning, analytics and computing solutions for plant assembly efficiency, vehicle quality and production flexibility. These include more efficient control of material flow, the early detection and elimination of supply bottlenecks and process disruptions, and the optimized operation of machinery and equipment in all plants. In addition, the cloud-based platform with its simplified data exchange is an essential prerequisite for Volkswagen to provide new technologies and innovations rapidly across its various locations. These include smart robotics, and data analysis functions to analyze and check shop floor processes from plant to plant. With the cloud-based platform, new applications, for example in IT-security for shop floor systems, can be scaled up direct to all locations throughout the world. Volkswagen will leverage AWS innovation best practices to become more agile and react faster on industry trends.
The architecture will be the new Digital Production Platform (DPP) from Volkswagen in future. All the Group’s plants and companies outside the Group will dock their system architectures onto this platform. This platform will standardize and simplify data exchange between systems and plants. “This is key step to enabling the type of comprehensive analytics that fuel maintenance and anything else a progressive manufacturer can get out of data,” says Vogels.
For Vogels, continuing to expand machine learning, IoT and edge computing will remain key focus areas for his team. However, security is the top priority. And it is also an area where his team has worked to be innovative. “As we work to bring legacy devices online in one form or another – whether it is into a private network, into the cloud or into a private network linked to the cloud – we need to recognize that this data is crucial to the manufacturing organization. It is industrial gold. We need to make sure that our customers are protected there. Now more than ever security needs to be the top priority,” he says.
A big piece of the puzzle? Building new tools that enable manufacturers to take actions to help themselves, explains Vogels. “Machine learning will play an increasingly important role here. You cannot build one service that works for all of your customers, especially when it comes to security. You need to be able to learn from the environment.”
Focus on education
Technology is almost always being used for a particular task – no one is creating technology for technology sake. Yet talent is lacking for most companies and continues to be the biggest challenge. At the same time, most companies have decided that cloud is the future – whether they are using it for analytics, product development or to dynamically control the entire supply chain. It is interesting to watch companies who have traditionally been slow moving, because of the massive investments they have into their assets, are now are moving full force into the digital domain, explains Vogels.
“They are managing a completely new world for them, and in many cases, without any access to digital talent. As cloud technology slightly settles down, the focus is now on how can we train people fast enough, so that all of these companies that are either going through or need to go through the digital transformation actually have access to the talent that they need,” he says. “We are looking forward to helping establish educational programs.”
Vogels is passionate about helping manufacturers successfully go through the digital transformation. After all, that’s where the real future of manufacturing thrives.
“Giving manufacturers the right tools, so that they can focus on what they actually want to do with their data as they continue to optimize operations,” he says. “This is true whether it’s Siemens Healthcare building machinery that has IIoT fully integrated or a smart manufacturer who still needs to be able to move back and forth between the cloud and on-prem. We need to make sure that we are building technologies that they can use five to 10 years from now because that is the process manufacturers go through.”