作者: Florian Meyer , Henk Wymeersch , Markus Frohle , Franz Hlawatsch
DOI: 10.1109/JSAC.2015.2430519
关键词:
摘要: We introduce a distributed cooperative framework and method for Bayesian estimation control in decentralized agent networks. Our combines joint of time-varying global local states with information-seeking optimizing the behavior agents. It is suited to nonlinear non-Gaussian problems and, particular, location-aware For estimation, combination belief propagation message passing consensus used. control, negative posterior entropy all maximized via gradient ascent. The layer provides probabilistic information form sample representations probability distributions. Simulation results demonstrate intelligent agents excellent performance simultaneous self-localization target tracking problem. In localization scenario only one anchor, mobile can localize themselves after short time an accuracy that higher than performed distance measurements.