Decentralized Bayesian Multi Agent Multi Target Search, Localization and Tracking
Abstract
We propose to develop new high performance algorithms for unsolved problems in target search, tracking and localization. We consider both centralized and decentralized settings, with multiple airborne mobile sensors and multiple stochastic moving targets. We focus on resource allocation and path planning to best learn target locations in settings where uncertainty plays a prominent role. How should sensors allocate bandwidth and battery resources toward communication and movement? How should sensors position themselves in light of incoming information about the target and other sensors? We plan to solve these problems within a Bayesian team decision framework using tractable computational methods that are robust to uncertainty about system dynamics and targets’ behavior.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 14, 2022
- Source ID
- FA95501910283
Entities
People
- Peter Frazier
Organizations
- Air Force Office of Scientific Research
- Cornell University
- United States Air Force