Non-Coherent RADAR for Intent Tracking
There is a great need for wide-spread, cost-effective perimeter monitoring for military, civilian, corporate, and national installations. While the use of non-coherent (marine) radar units makes the solution cost effective, the processing of the data is much more complicated. Traditional radar filtering techniques approach the problem of improving the range and velocity estimates of each target.
This work is concerned with the problem of finding and tracking humans using a stationary radar in a non-urban environment. We assume that the continuous radar measurements have been processed and a detection threshold has been applied to produce a set of contacts. The contacts themselves are indistinguishable. Our primary objectives are to identify which contacts originate from human targets and to associate those contacts generated by a single target. In addition, we want to be able to accurately predict the future position of any target under track. A track is defined here as the set of contacts associated with a single target which describes the path in space and time traveled by the target.
On the surface this is a typical multiple target tracking (MTT) problem. A variable number of targets are moving continuously in a region and the path of each target must be determined based on a sequence of noisy position measurements. However, the stated objectives along with the nature of human motion and the chosen sensor conspire to render traditional approaches to MTT inappropriate for this problem.
multimedia
Publications
-
(1)n/a
People
-
• Gabriel Elkaim, Associate Professor, Computer Engineering, UCSC, 831.459.3054
-
• Eric Poblenz, Ph.D. Student, Computer Engineering, UCSC, 831.459.2140
