Navigation of a mobile robot using reduced particles based on incorporating particle filter with neural networks

This paper proposes an enhanced navigation algorithm using reduced particles based on incorporating particle filter with neural networks. Though in global localization, the Particle filter is useful, it needs so many particles for increasing accuracy. Increasing number of particles increases computational burden. The proposed approach which incorporates PF and the NN (RBF) increases accuracy on navigating with smaller number of particles. The PF fused with RBF for training the weights. This process operates before resampling step of the PF. We confirm the result as comparing some algorithms including the proposed RNPF. The simulation results show the enhancement of bobust navigation through the training weights.