An improved particle filter algorithm is proposed for WSN-aided robot localization. This algorithm introduces node reliability and proposes its expression to realize information fusion among nodes. After having computed node reliability, sensor nodes with high localization, and take effect on updating particles according to their reliability. Furthermore, a novel simulation system for WSN-robot based on USARSim is developed, which is used to test our algorithm in this paper. The simulation result shows the good performance of the proposed algorithm in precision of robot localization and convergence speed of particle filter, and the conformity of the simulation system to the requirement of WSN-robot system research.