Bias Removal Techniques for the Target-Object Mapping Problem

• Main Author: Humke, C J
• Format: Report
• Language: English
• Subjects: ALGORITHMS; DATA FUSION; FIRE CONTROL RADAR; HOMING DEVICES; INTERCEPTORS; MAPPING; MAXIMUM LIKELIHOOD ESTIMATION; MULTISENSORS; RADAR TRACKING; SCENARIOS; Target Direction, Range and Position Finding

Multisensor track and data fusion require accurate sensor registration and positional bias removal. This report presents the problem of bias removal in the context of midcourse interceptor engagements. For this application two sensors are employed: a fire-control radar and an interceptor with an optical seeker. The fire-control radar tracks objects, collects data on the objects, performs discrimination, and hands over a target object map to the interceptor. Positional bias is generally evident in the target object map and must be removed in order to fuse the radar and seeker data accurately. Equations and algorithms are derived for calculating the maximum-likelihood bias estimate for three scenarios, given three assumptions about the measurement errors on the radar- and seeker-tracked objects. First, bias is shown to be an insignificant contributor to the assignment of objects when both sensors track the same objects and the measurement errors are equal and uncorrelated for all objects. Second, equations for direct calculation of the optimal bias are derived for cases where one sensor tracks a subset of the objects tracked by the other sensor. Third, algorithms are discussed for the most complicated scenarios, those where dissimilar objects are tracked by the two sensors.