A new tool applied to robot perception by selecting evidence sources

In this paper, considering the imperfect (incomplete, imprecise, uncertain, inconsistent) information sources in information fusion, To avoid and reduce the influence of the mistaken or misleading information on the final precision of fusion, the ESMS (Evidence Supporting Measure of Similarity) filter is designed and added to generalized fusion machine (GFM), and an ESMS function (not unique) to select evidence sources before the multi-sources information entering into the Fusion Machine is proposed and proved. Due to considering the barycentre of belief masses in four cases respectively, the fusion result becomes more precise. At last, we apply the new tool coupling with the Fusion Machine based on DSmT and PCR5 to robot’s robot perception. It shows a better performance in solving the challenging work with ESMS filter than that without one.