Improvement of Computational Efficiency of Unscented Particle Filter by Automatically Adjusting the Number of Particles

Authors
Kenta Hidaka, Takuo Suzuki, Kunikazu Kobayashi
Corresponding Author
Kenta Hidaka
Available Online 1 September 2016.
DOI
https://doi.org/10.2991/jrnal.2016.3.2.14
Keywords
RoboCup, Self-localization, Unscented particle filter, Kalman filter, Particle
Abstract
In RoboCup Standard Platform League (SPL), the method using unscented particle filter (UPF) has been proposed for self-localization. The UPF resolves a problem of particle filter which cannot be sampled appropriately when the likelihood is too high or low. This filter can estimate accurate position when the more number of particles is. However, the more, the more computation time is needed. In the present paper, we propose an automatic adjustment method for the number of particles in UPF. The proposed method uses three kinds of feature values with respect to particles, i.e. centroid, standard deviation, and weight. Through computer simulations, we confirmed the improvement of computational efficiency of UPF?

Copyright
© 2013, the Authors. Published by ALife Robotics Corp. Ltd.
Open Access
This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)