Use of window addressable memories for high speed geometrical analysis

Abstract

Many real-time problems require assessment of the location of a moving platform with respect to a fixed or time varying set of objects. Examples of such problems are robots performing precision movement in a dense object space or low flying aircraft over enemy territory. The method described in this paper permits quick assessment of relative locations to two- or three-dimensional rectangular shaped objects with respect to an arbitrary number of points and subsequently to a real or fictive line of sight. The relative location assessment can be performed at a rate of up to 5 ns/data point/object and the line intersect evaluation at 5---500 ns/line/object dependent upon the geometry. Both the relative assessment algorithm and the intersect algorithm have been implemented in a small hardware unit called the TIGER, Three-dimensional Intersect & Geometrical Evaluation in Real-time, that can be incorporated in unmanned as well as manned systems in space, under water, and avionics. All quoted performance data is based upon analysis in three dimensional Cartesian space unless otherwise stated.