Adaptive High Dimensional Data Fusion and Sensing for Dynamic Target Detection and Tracking
Abstract
For cases with training data, the PI aims to develop joint nonlinear manifold learning techniques to estimate the manifolds or the mapping from the target state to the high dimensional heterogeneous observations. With the learned manifolds as observation models, multi-target tracking filters will be developed to fuse sensor data and track multiple objects in the surveillance region. Another PI’s goal is to investigate joint sequential detection and tracking, a challenging problem due to data dependence over time. The PI will derive the optimum decision structure for joint sequential detection and tracking with general data dependence structures. Theories and algorithms will also be provided for joint quickest detection and tracking and sequential detection of multiple dynamic objects. The PI plans to develop an adaptive UAV sensing framework to achieve optimum detection and tracking performance. UAVs’ coordination and collaboration will be guided by feedback from the sequential detector and object tracker. The objective function will be based on the Kullback-Leibler divergence or Chernoff information for detection, the conditional posterior Cramer-Rao lower bound for tracking, or their combinations for joint detection and tracking. Further, the PI proposes to develop non-myopic adaptive UAV sensing approaches.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jun 11, 2018
- Source ID
- FA95501810362
Entities
People
- Ruixin Niu
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- Virginia Commonwealth University