The sensing and perception requirements for NASA's automated operations are addressed in this project with a novel combination of tracking targets and algorithms that would simultaneously identify, locate, and track many objects viewed by a pair of color video cameras. The key innovation is the use of pre-affixed, multi-colored tracking targets and a hardware preprocessor for rapidly identifying them under variable illumination. This system offers advantages in capacity, speed, ease of programming, and cost. Hundreds of thousands of objects could be identified, and dozens tracked simultaneously. Locations and orientations of dozens of identified objects could be determined every 60th of a second. The system design is inherently self-teaching and requires only two low-cost, color cameras and one board built with off-the-shelf LSI and VLSI components. The anticipated positional accuracy is plus or minus several millimeters at a range of 10 meters. In Phase I, the company will build and test a prototype which has all the capabilities of the final system except real-time tracking.
Potential Commercial Applications: Applications would be found in numerous robotic vision systems for aerospace and industrial applications.STATUS: Phase I Only
