From the Small Picture to the Big Picture
UNM researcher applies fuzzy logic to sharpen research results
In the future, your family photo album could be filled with pictures that are perfectly lit, crystal clear, and accurately reflect your vacation to Hawaii, with its deep blue water, crashing waves, and vibrant green palm trees. You'll have Mo Jamshidi to thank for that.
And when NASA lands a person on Mars, Mo Jamshidi will deserve some thanks then, too.
That’s because as Regents Professor of Electrical and Computer Engineering, A&T Professor of Manufacturing Engineering, and Director of the NASA-funded Autonomous Control Engineering Center, Mo Jamshidi has conducted research that has literally ranged from looking at the small picture to the big picture. In the process, he has gained international recognition for his research, prolific writing, and leadership in the area of large-scale complex systems.
Some of Jamshidi's hard and fast research has its roots in something decidedly ambiguous called fuzzy logic. Fuzzy logic is a technology that applies human intelligence and reasoning, processes which are inherently imprecise, to machines. The tools apply a mathematical framework to human reasoning, replacing human thinking with "if-then" rules. The result is a decision-making process, or controller, that is then implemented on a chip and put inside a machine.
Fuzzy logic's strength is its ability to tolerate inaccuracies and shades of gray, whereas conventional "hard computing" handles only absolutes.
Fuzzy logic was first created by Jamshidi's longtime friend and mentor, Lofti A. Zadeh, a professor at the University of California at Berkeley, who pioneered the concept with a groundbreaking and seminal paper on fuzzy sets in 1965.
Since then, fuzzy logic has been integrated into a spectrum of products manufactured around the world, but dominated by Japanese corporations, which adopted the concept more readily than American manufacturers. Fuzzy logic is used for the motion stabilizer in a video camera, determines the best air-fuel mixture in car engines, provides a quiet and smooth ride on an elevator, and helps air conditioners run more efficiently. "Fuzzy logic can replace any kind of application where humans are in the loop and they don’t need to be," says Jamshidi.
The Small Picture
In the summer of 1992, Jamshidi, along with two UNM colleagues, was teaching a short course on fuzzy logic at MIT Lincoln Laboratory when some engineers from Polaroid Corporation approached him. They wondered whether fuzzy logic could help them solve a dilemma they were having with a video printer. They depended on an expert to fine-tune the tint, color, contrast, and sharpness of a picture before it was printed. But the printed picture wasn't matching the parameters set by the expert. Could fuzzy logic be used to correct the error?
Yes, Jamshidi thought. It was a perfect problem for fuzzy logic, so he secured a $50,000 grant from Polaroid and worked with a team of researchers and students to resolve the problem. The result was a fuzzy logic controller implemented on a micro-controller board. This controller applied "if-then" rules to automatically adjust any digital or analog picture's sharpness, contrast, tint, and color, pixel by pixel, until the print was perfect. UNM's Science and Technology Corporation (STC) patented the Fuzzy Logic Controlled Printer in 1996.
The Future of SmartPhotoLab®
The software by-product of this patent was a C++-based environment called SmartPhotoLab. UNM researchers have used it to refine satellite images for more accurate analysis of vegetation growth and soil erosion. Jamshidi says it could also be used in military applications to refine reconnaissance data or verify nuclear weapons test bans. STC is collaborating with Jamshidi and his students to commercialize SmartPhotoLab. Jamshidi says that while there's even greater commercial potential for SmartPhotoLab than for the printer, that's not what is important to him. "This patent and its by-product give outsiders a new perspective on the quality of the engineering work that's going on at UNM," he said.
The Big Picture
Fuzzy logic is just one tool Jamshidi uses in his major area of research called large scale systems or complex systems. He’s been researching the topic since he was a doctoral student at the University of Illinois in the 1960s, and is now recognized worldwide for his work in the field. In fact, Jamshidi wrote the first textbook on large-scale systems in 1983. The book and its second edition have been translated into five languages and are used at universities around the world.
A system is considered to be large scale if its size, design, model, and function are so complex and decentralized that standard analysis, modeling, and computing procedures are ineffective for managing and troubleshooting the system. You'll find large scale systems everywhere - in power grids, hospitals, and economies. Jamshidi's research uses, among other things, fuzzy logic and artificial intelligence to solve problems associated with large-scale systems and help them run more efficiently.
The Biggest Picture
Jamshidi's newest area of study, System of Systems (SoS), is a natural extension of his large-scale systems research. SoS focuses on the integration of multiple large scale systems that operate independently but also work interdependently within a larger system, or metasystem. For example, an airplane with its many automated parts and computer-controlled mechanisms is a system. But an airport with hundreds of airplanes, gates, control towers, and administrative departments is a system of systems. The military, satellite systems, and even large corporations are other metasystems. "Large-scale complex systems are the building blocks of SoS," says Jamshidi.
The challenges that researchers face with large-scale systems grow exponentially with metasystems. "The question is that, even though each system works fine individually, how will all the systems in the system be able to work together with some pre-specified degree of functionality and efficiency?" explains Jamshidi. SoS research looks at the best way to allocate resources correctly and integrate the many individual systems so that the metasystem functions effectively with minimal probability of failure.
Fuzzy logic comes into play in most stages of SoS. The soft computing process - which includes fuzzy logic, neural networks, evolutionary computing, and probabilistic reasoning - is useful in modeling the problem, aiding in information exchange between each system, and in analyzing problems within the system.
SoS was initiated by the Department of Defense and the U.S. aerospace industry prior to 9/11. Since then, the idea has been adopted by all branches of the military, major defense contractors, NASA, and aerospace companies. "SoS has a wide range of applications," says Jamshidi. "I think it's potentially a $2 billion per year opportunity for research and development." In fact NASA and CalTech's Jet Propulsion Laboratory in Pasadena are using SoS on the Code T Project, President Bush's initiative to send a person to Mars via the moon.
SoS is also being used in Global Earth Observing Systems (GEOS), a 60-nation global effort led by the United States government and the aerospace industry.
While SoS is gradually being adopted, research in the field is just beginning. Jamshidi plans to take the lead at UNM. He taught UNM's first SoS course in the spring of 2005 and is planning to write the first textbook on the topic.
Jamshidi is also organizing the National Consortium on System of Systems, a high-profile think tank of SoS experts from academia, government agencies, national laboratories, and industry. The consortium being planned by Jamshidi will collaborate to solve many of the most challenging problems facing the country and the world community today. Jamshidi plans to hold the first consortium and IEEE conference next spring in Los Angeles. "We are combining talents from different places so that a theory developed at the university level could be implemented for homeland security by a defense contractor. We're all working on pieces of the same puzzle," he said.
In essence, Jamshidi is using a system of systems approach to solve the challenges of system of systems. And based on his track record, the results will be anything but fuzzy.