A novel particle filter named chaos map sampling particle filter(CMS-PF) is proposed to solve the particle impoverishment problem. After important sampling, chaos series obtained offline are transformed into some sub-spaces, whose kernels are the particles with heavy weights, by using the algorithm similar to carrier wave to generate map particles. The map particles are associated with predictive particles to construct the candidate article set and the optimizing selection of particles is realized based on its own weighs. Simulation results show that this method can effectively improve the state estimation precision.