Virtually every solid-state device used in an industrial application has some type of real-time operating system (OS). Low-level devices, such as sensors and transmitters, usually have a proprietary OS, and high-level devices, such as PCs and PLCs, usually rely on an off-the-shelf OS.
There are several ongoing trends in the commercial OS market that will certainly affect industrial machine control systems. Chief among these trends is distributed computing, which will allow an application to run on multiple networked computers.
"Distributed computing advances will enable control applications that run on different devices to appear to operate on a common virtual platform with varying degrees of cooperation," predicts Raj Govindaraj, software architecture principal engineer with Rockwell Automation (www.rockwellautomation.com). "This will allow developers to think in terms of the control problem they have to solve rather than how to distribute their application across multiple control system components."
According to Norma Dorst, software marketing manager for National Instruments (www.ni.com), a distributed computing OS will allow software developers to partition processors. A software developer will be able to create two 50 MHz processors from a single 100 MHz processor. The OS will provide infrastructure for each virtual processor to be programmed independently.
Operating systems also will enable networks of synchronized distributed processors. "It will be possible to assemble processors into a distributed control system that acts more intelligently at the point of control, often delivering real performance improvements," adds Dorst. "In addition, distributing the intelligence makes adapting and adding functionality to the control system easier."
Distributed computing will certainly affect machine control systems. The first step will come as industrial control software firms modify applications to run on an OS that takes full advantage of distributed computing. Such an OS will recognize every intelligent device in the control system, and it will use the resources of each device as needed.
Vendors that now market real-time PC-based control software would switch to selling real-time control software designed to run on a variety of OSes and hardware platforms.
A machine builder might purchase real-time control software conforming to the IEC 61131-3 standard. With current OSes, application development is constrained by the target hardware. Machine builders must be aware of computing system resources and network architectures.
In the future, the developer will specify a performance requirement such as execution speed. The OS will distribute computing tasks among multiple computing devices to meet the requirement. The distribution of computing tasks will not be static--it will be continuously adjusted by the operating system to optimize performance and reliability.
Available computing devices might include HMIs, PCs, PLCs, transmitters, servo drives, or sensors, all networked together. If a machine builder wanted to increase performance, another device could simply be added to the system. The operating system would recognize this new device on a plug-and-play level and would immediately distribute computing tasks to the device.
Reliability surely would be increased. The addition of another computing device with sufficient system resources could allow the control system to operate in a fully redundant mode. Redundancy would be under the purview of the OS and would be automatically provided as allowed by total system resources.
Because many devices will be expected to simultaneously execute multiple applications, a technique called dynamic program loading will be required. "Dynamic program loading enables software to be changed in a system that is running," says Adrian Leufven, director of product marketing for OSE Systems (www.ose.com). "New services can be added, old services can be deleted, software can be upgraded, and parts of a system can be reconfigured without stopping the system."
Like the commercial computing world, industrial operating systems will become cheaper and faster. This means a commercially available OS will be embedded in a host of applications that now require a proprietary OS and task-specific electronics.
"Traditional controller/sensor/actuator distinctions will blur because there won't be a financial barrier to incorporating functionality into smaller, more mundane devices," predicts Mike Nager, industry marketing manager for Phoenix Contact (www.phoenixcon.com). "If nanotechnology develops as some have proposed, we could have entire machines--microscopic in size, each with its own distributed OS."
Others predict that embedded control and monitoring systems will be everywhere, and that each of these embedded systems will have its own low-cost, high-performance, and highly reliable OS. And I do mean everywhere: "Your body will have an embedded controller to monitor and control insulin levels, to monitor blood pressure, and to detect cancer and other diseases," says Dan O'Dowd, founder and president of Green Hill Software (www.ghs.com).
E-mail Dan at firstname.lastname@example.org.