THE OBJECTS OF MY AFFECTION
To discuss an alternative approach to factory control software, we again have to talk of real-time computing. A real-time system consist of a controlling system and a control system. The controlling system interact with the environment of which the controlled system is a part. Severe consequences result if the timing and logical correctness of the system are not satisfied.
If we want to look into the future, we need to talk about the latest philosophies. One of the newer approaches to software programming is the object-oriented programming system (OOPS), which has already been very successfully applied in conventional computer systems. Object oriented technology is both strong and mature in the database and graphics arenas. With OOPS on the real-time horizon, the grail of interchangeable, reusable and updatable code comes tantalizingly into view.
What is OOPS? Think of a Russian egg. You know, Fabergé eggs within the eggs that the children of the Czar got for Easter. Well, an OOPS uses software eggs , or boxes, that contain within them both the code and the data. The box is called an object. An object can be a single data point or line of code, or it can be made up of complex set of data and code. The gist of it is that the objects are data-centered, with the code subservient.
There can be many types of objects, including graphical, database and programming units. Objects are both a set of operations and memory "owned" by the object. Conventional languages support a data procedure model in "one long string." With an OOPS, the code and data are treated as blocks, not strings.
Why all the excitement about the OOPS? The use of objects greatly reduces the cost of programming a system. And the technique can be practiced on existing systems. The boxes can be big or small, simple or complex. For dealing with relatively simple records an numerical data, a relational database model may be the best. But for symbols and interwoven data, objects work better. This makes objects ideal for user interfaces. One could also imagine ladder logic programs and data used as objects in PLCs.
But factories today do not use OOPSs to control processes. PLCs do not use a distributed objects system in plant control. No sensor-based objects are in wide use. Why? Because the developers of technology did not try to adapt OOPSs to the needs of real-time systems. However Digital Equipment Cop. and Hewlett-Packard are working with objects in the context of real-time control systems. We all suspect that the leading PLC suppliers are working on versions of real-time OOPSs.
Let's talk about implementation. The technology should be able to make links from PLC to PC to minicomputer transparent, i.e., the user will not be aware of the techniques used to effect the links. The type of I/O modules involved should not matter either. Interface standards for non-real-time applications that address the issue of object definitions, methods and messaging are due by 1994. Some feel that truly complex systems will not be possible without such standards.
How should the complex systems work? Don't know. However, we can make some guesses. Assume open hardware baskets of I/O. Roaming eggs of data and code wander around and lock onto jobs. Each job is an application of a module in a distributed system. The control workstation can operate with an object anywhere in the system. The workstations a single point of control. The distributed control system is transparent to the workstation and operates via the objects resident in the modules. Objects must be able to reconcile sensor data with the internal data/ code.
Problems still around. No sensor concepts for objects are technically mature. Yet the problem of reconciling sensor data in large control systems has not been sufficiently explored. Does sensor and embedded data needs to be ergodic? Issues is messaging, sensing and operating systems need to be resolved. Language and data representation are not an issue. AS usual, culture is the true impediment to utilization.
Objects can integrate and optimize with standard products.
Objects can interface to a complex heterogeneous system.
Objects are independent of location and reusable.
The first vendor to attack and solve the control issues using objects will be the king of silicon hill. To paraphrase Einstein, "thinking is concepts (objects), interconnection, and the rationalization of both with the senses." We can expect to see OOPS-based products in 1995 and wide use by 2002. The PLC of the future is software objects of reusable code mounted on open systems.
As appeared in Manufacturing Systems Magazine January 1993 Page 56
References - Table of Contents
|R. Morley Incorporated
614 Nashua Street, Suite 56
Milford, NH 03055-4992 USA
Tel: 603-878-4365 FAX: 603-878-4385