Dynamic Data Driven Electro-Optical Sensor Detection, Tracking and Multi-Objective Control
Abstract
A basic research effort using a Dynamic Data-Driven Application System (DDDAS) methodologyto improve the detection, tracking, and characterization of objects using Electro-Optical (EO)sensors in Space and Air Domain Awareness (DA) applications is proposed. Domain Awarenessis the actionable knowledge required to predict, avoid, deter, operate through, recover from, andattribute cause to the loss or degradation of space and air assets. For EO sensor detection, tracking, and classification of aircraft (i.e. commercial, UAVs, military) and space objects (i.e. debris, satellites, launch vehicles), current systems do not dynamically select and improve upon models in real time, and largely generate results in post-processing. Much relevant data is not captured in-situ, observations of opportunity are missed, and followup opportunities are squandered. Thiscan result in inefficient asset utilization or loss of space assets. Proposed basic research includes1) A Markov Modulated Poisson Process brightness model to enable model selection betweenfull Bayesian ‘detectionless’ object detection and tracking and classical moving target indicatorbased multiple hypothesis tracking approaches, 2) an observability-driven Dempster-Shafer (DS) approach to full Newtonian dynamics identification and classification, and 3) a real-time EO sensor multi-objective optimal control policy to collect data to inform object existence, classification, characterization hypotheses, and full state estimation.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 28, 2017
- Source ID
- FA95501710209
Entities
People
- Marcus Holzinger
Organizations
- Air Force Office of Scientific Research
- Georgia Tech Research Corporation
- United States Air Force