This award is to purchase a multi-processor host machine for the Maspar computer and real-time high-speed, image capture and display hardware in order to support research in parallel real-time, high-speed computer vision and visualization. The upgraded Maspar system will enable research to be conducted in the areas of artificial neural networks for robotics, factorization methods for shape and motion from image streams, road and terrain navigation, visualization techniques for relativistic imaging, and the development of a stereo vision machine. Research in parallel real-time computer vision and visualization is computationally expensive and requires both massively parallel computational power and real-time image I/O support. Carnegie Mellon University will be upgrading their Maspar computer and adding real-time image acquisition hardware to support research projects in these areas.