Robot visual servo with fuzzy particle filter

This paper proposes a robot visual servo method with an adaptive particle filter based on fuzzy logic theory to online estimate the total Jacobian matrix of a robot visual servo system. A set of fuzzy rules are used to select appropriate numbers of particles according to the filtering estimation errors. When an estimation error is high more particles are used, and when the estimation error is low fewer particles are used. The visual servo results on a two degree-of-freedom robot system show that the proposed fuzzy adaptive particle filter visual servo method needs less time than that of traditional particle filter visual servo method to get a comparative tracking accuracy.