Abstract: In recent years,there has been increasing interest in the development of robotic distributed sensor networks,but little is known in the area of the architecture and strategies for multisensor data fusion.This paper constructs a model of two-layer architecture and strategies of sensor fusion of the data level and evidence level.The model embodies the principle of information processing in data fusion problems.According to the proposed structure,this paper describes the efficient algorithms for multisensor data combination and information fusion usin Bayes parameter estimation and Dempster-Shafer's theory.An example is considered in detail.

Key words: sensor fusion, information fusion, estimation theory, uncertainty reasoning