An intuitive and efficient switching particle filter for real-time vision-based localization

The particle filter algorithm has been applied to mobile robot localization problem for more than a decade. In this paper, we present two extensions of the basic particle filter algorithm for humanoid robot vision-based localization in a soccer game. The individual particle reset module allows localization even in kidnapped robot problem. The “switching algorithm” is developed to save computing time through adapting the population size of particles. The accuracy and efficiency of the proposed approaches have been verified in a simulated humanoid soccer environment.