Suboptimal joint probabilistic data association
AssigneeInternational Business Machines Corporation
InventorsGary L. Phillis, James A. Roecker
Life of this patent
Estimating the term…
Abstract
A computer method for accurate target tracking by a faster method for hit-to-target data association. The method involves using an approximation of a weighted centroid of hits to update the calculated tracks. The result, while losing theoretical accuracy since it approximates the calculation, produces fast calculations. Further, the calculated result has sufficient accuracy for most practical applications.
Claims5
In a computer implemented target tracking system, including a plurality of sensor stations, operatively connected to a data processing system, a computer method for updating the tracks associated with multiple targets using hit-to-target data association, comprising the steps of:
- a) identifying a plurality of within said data processing system targets to be tracked by said plurality of sensor stations and creating an associated track for each of said targets, said track comprised of a collection of points in space, each of said points associated with a specific point in time;
- b) for each of said points in space, selecting an associated extension gate within said data processing system, said gate being an area centered about said point;
- c) at each of a sequential series of points in time, receiving a plurality of hits from said plurality of targets, said hits being received by at least one of said plurality of sensor stations, and said hit comprising a received signal from one of said targets;
- d) selecting a first of said targets to be updated and identifying said target's associated first track;
- e) selecting a first point in time from said sequential series of points in time;
- f) identifying a first point in space belonging to said first track, said first point associated with said first point in time;
- g) identifying a second point in time, said second point in time immediately preceding in said sequential series said first point in time;
- h) identifying a first extension gate, said first extension gate being associated with said first point;
- i) identifying each of said plurality of hits which were received at said first point in time and which lies in said first extension gate;
- j) selecting from said plurality of tracks, each of said tracks which enclose one of said identified hits;
- k) for each of said identified hits, calculating a weight;
- l) selecting a first hit from said identified hits and calculating a suboptimal probability for said first hit, said suboptimal probability obtained by dividing said weight associated with said first hit by the sum of a constant B and the sum of all of said calculated weights within said data processing system;
- m) calculating a value v(k) of said first track, by taking the sum of said suboptimal probability of each of said identified hits multiplied by a difference between said hit and said first point within said data processing system;
- and
- n) updating and displaying said track associated with said first target by replacing said first point by the sum of said point associated with said second point in time and v(k).
Dependencies are inferred from the claim text.