On Generalized Bayesian Data Fusion with Complex Models in Large Scale Networks
This work addresses robust information sharing for autonomous agents in sensor networks, representing an incremental improvement with specific algorithmic advances.
The paper tackles the challenge of implementing generalized Bayesian decentralized data fusion (DDF) for complex belief models in dynamic networks, and demonstrates that their novel algorithms lead to significantly better fusion results in multi-robot target search examples.
Recent advances in communications, mobile computing, and artificial intelligence have greatly expanded the application space of intelligent distributed sensor networks. This in turn motivates the development of generalized Bayesian decentralized data fusion (DDF) algorithms for robust and efficient information sharing among autonomous agents using probabilistic belief models. However, DDF is significantly challenging to implement for general real-world applications requiring the use of dynamic/ad hoc network topologies and complex belief models, such as Gaussian mixtures or hybrid Bayesian networks. To tackle these issues, we first discuss some new key mathematical insights about exact DDF and conservative approximations to DDF. These insights are then used to develop novel generalized DDF algorithms for complex beliefs based on mixture pdfs and conditional factors. Numerical examples motivated by multi-robot target search demonstrate that our methods lead to significantly better fusion results, and thus have great potential to enhance distributed intelligent reasoning in sensor networks.