• Home »
• Science & Technology

The machine control problem is normally approached from the perspective of having a central body of intelligence (and control) in the machine. However, we are developing a conceptual design of a machine using distributed learning and intelligence. This new design is loosely based on biological models of social insects. For example, in an ant colony each ant functions according to local rules of behavior. There is no "king" or "queen," although the latter name has been given to the reproducing ant. Following a similar approach, we are developing a modular machine architecture in which each machine element has local rules of behavior (and local learning) along with a global element that influences local behavior (but does not dictate actions). A prime goal is to develop methods of learning and behavior modification that ensure global stability and optimization of the total machine.

Consider an intelligent machine in which various machine functions are carried out in a distributed manner. In addition to the machine hardware required there are several "agents." These agents have local control of various machine functions and are able to communicate with each other and with an operator agent. Communications are carried out using a simple vocabulary; each agent has an on-board interpreter. The operator agent may be a human or an interface to a human (or even an interface to another machine). (Although it would be possible to focus on autonomous machines, we chose not to do so; our machines interact with humans who have supervisory control authority.) The various agents incorporate knowledge of how to perform tasks, the ability to learn from experience, and memory of past performance. The agents will eventually be able to optimize both their local behavior and the global behavior of the total machine.

Throughout the DOE complex and the INL, there is a wide range of highly nonlinear dynamical systems which need to be mathematically modeled as well as qualitatively understood. Such an understanding is the first step in controlling such processes. These complex systems include industrial processes (Welding), fuel processing (coal desulfurization), bio-remediation (TCE – degradation), transport through the vadose zone (water and contaminate flow), etc. We at the INL are investigating and developing new tools in the fields of chaotic mathematics and complexity theory in order to accomplish these goals. For more information about this research, access the Chaos and Complexity web site.

Previous research focused on welded processes used in a variety of industries such as oil and gas, chemical, shipbuilding, automotive and transportation, energy generation, aerospace, bridges, buildings, and consumer products. Weldments contribute significantly to the economic and strategic health of the nation. Welding is presently an industrial art based on the manual skills of the welder or very simple duplication of those skills by an automatic, but dumb machine. This project will establish a scientific basis for intelligent welding machines that incorporate both knowledge of welding physics and empirical learning capabilities.

Additional reading:

• The VIPER-SEW Project, Building a Robot that Slithers O.J. Schubert, T.A. Barnes, D.J. Hunsaker, R.A. Anderson, C.R. Tolle*, H.B. Smartt, K.S. Miller. — 1.2MB PDF
• Complex Intelligent Machines, Proceedings, Eighteenth Symposium in Energy Engineering Science, DOE-BES, Argonne National Laboratory, May 15-16, 2000, Herschel B. Smartt, Charles R. Tolle, Kevin L. Kenney. — 146kB PDF
• Intelligent Control of Modular Robotic Welding Cell, 6th International Conference on Trends in Welding Research, 15-19 April 2002, Herschel B. Smartt, Kevin L. Kenney, Charles R. Tolle. — 35kB PDF