A particle filter resampling method based on improved genetic algorithm
Rong Zhou, Menghua Wu, Kemin Zhou, Jing Teng · 2017
Particle impoverishment, which always leads to accuracy decreasing, is a classical problem of the traditional particle filter resampling. A particle filter resampling method based on improved genetic algorithm is proposed in this paper to solve the problem. In the improved algorithm, a genetic resampling step is applied for particle reproduction. The main work of the improved algorithm include two parts. The first one is calculating the dynamically changed crossover probability and mutation probability which is used to decide whether the genetic resampling step will be carried out. The second part is adopting the Euclidean distances of predicted particles as standard of the crossover and mutation operation for its low computational cost and effectiveness. Simulation results show that the proposed method has better compromises between the computational overhead and the tracking accuracy, compared to the traditional resampling algorithms.