A coordinated, interdisciplinary program of experimental research and analysis for the development of a smart, computer controlled optical sensor for processor-controlled applications is proposed. Spatially resolved measurements of temperature and gas composition in process streams are made, using multiple line-of- sight infrared absorption measurements and tomographic inversion schemes. The research aims at progress in the area of suitable inversion methods, their adaptation to vector parallel processing, signal-to-noise improvement and system integration for adaptive control. Such an on-line adaptive control would allow for the practical automatic optimization of powerplant operation in conformity with maximum allowed pollutant levels. Since the regulatory constraints on this optimization are independent of the choice of fuel or combustor, the algorithm will be able to search for an optimum figure of merit, regardless of the grade of coal delivered to the plant. In other words, the process is not prededicated to a certain type of combustion; it is truly adaptive. The expected savings in coal consumption would have large economic proportions.