Based in Austin, Texas, National Instruments (NI) has a long history of creating electronic testing and embedded control solutions designed to make data-collection and wireless-sensor systems operate accurately and efficiently. Today, the company has $1.2 billion in annual revenue and a wide-ranging customer base (including Airbus) that is implementing industrial IoT solutions. The firm also encourages its own employees to innovate and experiment: Brad Hughes, a software engineer at National Instruments, led a team that won an Intel-sponsored hackathon last fall, for developing an Internet-linked food scale. To learn about the company’s approach and how the IoT is changing the modern factory, IOT Journal spoke with Jamie Smith, the firm’s director of embedded systems product marketing.
IOT Journal: Can you explain the basic concept of the industrial Internet of Things and how it differs from legacy industrial systems?
Jamie Smith: The concepts of the IoT are similar between the consumer IoT, where we see products such as Fitbit, smartphones and even smart cars, and the industrial IoT or IIoT—which is also called, simply, the Industrial Internet. Both are characterized by dispersed, sensor-based devices that communicate to a central hub. But the technology requirements for consumer IoT and the IIoT are different.
Traditionally, an industrial system worked by using one central controller, or brain, which communicated out to sensors or I/O [input/output] devices. Now, those sensors are getting smarter and more autonomous. Also, those smart [sensor and controller] systems are increasingly out at the edge [of the industrial architecture]. They’re not just doing simple things but taking on a broader range of responsibilities to make decisions, improve insights of factory workers, and doing overall better productivity.
Another key to the IIoT is the use of field-programmable gate arrays (FPGAs), which are chips that can be programmed by the end user—as opposed to chips used in, say, personal computers, which come with an installed operating system or application.
In a traditional industrial system, sensor data comes in through a data bus, and that data bus would control an actuator. That round trip takes 100 milliseconds, or as little as 10 milliseconds. But using an FPGA, we can have that round trip down to the microsecond timeframe.
Traditionally, chips were designed in software, and prototypes were made in FPGAs. They were just a mechanism to get to what you were making. Now, FPGAs are smaller, more power-efficient and integrate more elements than just gates [which control flow of electrical current in a chip], memory and interconnects. They also have digital signal processors so now you can create custom reprogrammable chips that do whatever you need them to do. So you can create functionality in hardware without needing an operating system and without software bugs or conflicts.
IOT Journal: National Instruments helps companies deploy IIoT using its LabVIEW Reconfigurable I/O (RIO) platform, which includes an FPGA that the end user can configure using LabVIEW software. How does this approach make NI unique?
Smith: Our RIO devices combine a microprocessor and an FPGA, and the end user uses this to operate off-the-shelf measurement or control hardware.
Most I/O systems have an FPGA, but those FPGAs are usually closed and can’t be set up [by the end user]. They handle the I/O and pass it to a processor, where software makes the decisions and changes. To program FPGAs on your own, you would need a digital designer, and this is a specialized job—there are maybe 1,000 software developers for every FPGA.
For example, think of very fast control loops, on a factory floor. A company like Airbus, that is controlling a motor or control system… that system would be running on dedicated hardware. But [in the past] if the customer wanted to use the same control system in a brand-new style of actuator [for a different factory application], they would need to redesign the hardware. [With LabVIEW] they can use [our] same piece of electronics and reconfigure it without having to redesign the hardware.
Another example [of how LabVIEW can be used] might be a piece of electronics that is measuring something on electrical equipment and is distributed on the grid. It operates based on a standard, but if that standard changes, the utility would have to replace their old measurement equipment. But [with NI FPGAs] we can just redesign the product from the control center. So [for the customer] it provides future-proofing. The LabVIEW software gives flexibility, because you’re using software to design the way your hardware operates.
IOT Journal: Can you tell me more about NI’s work with Airbus?
Smith: Airbus wants to improve its overall performance through its project called the Factory of the Future. They want back-end systems that integrate tools and autonomous robotics systems being used alongside workers, and to do that, they need [systems] to be able to communicate with each other. They asked: “How do we make the most optimized system to perform each task?”
Initially, they developed optimized hand tools, robotics, etc. But when they tried to integrate them, it was cumbersome. They took a step back and said, “Let’s use the same architecture and much of the same software.” They used our RIO architecture on wearables, smart tools and autonomous systems, and allow those to communicate with back-end systems. The overall development time dropped 90 percent.
IOT Journal: How does adding intelligence to the tools and autonomous systems and linking them to the RIO architecture help?
Smith: Inside a plane’s fuselage, there are more than 400,000 fasteners and more than 1,000 tools used to drill and tighten them. The tools communicate wirelessly to the back-end systems. Workers are issued wearable devices that also communicate with the back end to confirm that they are using the tools optimally—putting the right amount of force on each fastener, for example.
We characterize the Airbus Factory of the Future as a cyber-physical system, which operates on its own but is connected to back-end systems and other machines near it. Over time, the system learns about its behavior, sharing information about itself to a back-end system. The machines can say, “I’m performing in this way, and I’m also going to tell you I believe I’m getting close to breaking down.”
There are multiple layers to the industrial IoT. One layer is comprised of machines that are aware of their surroundings and how they operate, such as a machine that tracks, over time, how much power it needs to operate. If it takes more energy over time, it might be because of wear. Or maybe windings are getting damaged so it takes more current. That is one way a machine can learn.
Another way is by measuring and tracking vibrations and sounds. There, you get into more advanced analytics of matching the characteristics of a vibration to a part that is worn out or broken down mechanically. This is not that different than how a car’s air pressure tire monitor operates, though that is a very simple example.