A Model for Target Tracking in the Presence of Geographic Constraints.
Abstract
For accurate tactical scene estimation, automated target tracking algorithms must take into account environmental information, often in the form of kinematic constraints. This need is particularly important for waterborne targets operating in littoral waters, where the presence of geographic constraints (e.g., landmasses) can strongly bias predicted target motion. Such constraints are generally modeled through the use of numerically intensive techniques that cannot be readily applied to multitarget, multisensor scenarios. This report presents an alternative modeling technique that approximates the inadmissible region (the landmass) using a sum of analytic functions. A non-homogeneous process noise field is employed to model the uncertainty associated with a target's evasive maneuver as it approaches the constraint boundary. This non-homogeneous process noise field is used in conjunction with standard Kalman filter equations to predict and track targets in the vicinity of the constraint; a windowed Gaussian probability density is used to describe estimation uncertainty. Results for scaled simulations are presented.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 16, 1996
- Accession Number
- ADA310143
Entities
People
- M. L. Graham
Organizations
- Naval Undersea Warfare Center