C38. Appraisal of Different Particle Filter Resampling Schemes Effect in Robot Localization

This paper considers the effect of the Resampling schemes in the behavior of Particle Filter (PF) based robot localizer. The investigated schemes are Multinomial Resampling, Residual Resampling, Residual Systematic Resampling, Stratified Resampling and Systematic Resampling. An algorithm is built in Matlab environment to host these schemes. The performances are evaluated in terms of computational complexity and error from ground truth and the results are reported. The results showed that the localization plan which adopts the Systematic or Stratified Resampling scheme achieves higher accuracy localization while decreasing consumed computational time. However, the difference is not significant. Moreover, a particle excitation strategy is proposed. This strategy achieved significant improvement in the behavior of PF based robot localization