Mobile Robot Localization Based on Gradient Propagation Particle Filter Network

In order to solve the problem that the gradient information can’t propagate backward due to the non-differentiability of resampling process in the end-to-end training of Differentiable Particle Filters (DPFs) network model, a particle filter network with gradient propagation is proposed i...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Heng Zhang, Jiemao Wen, Yanli Liu, Wenqing Luo, Naixue Xiong
Formato: article
Lenguaje:EN
Publicado: IEEE 2020
Materias:
Acceso en línea:https://doaj.org/article/063ce98188f84191bb67566055a282ff
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:063ce98188f84191bb67566055a282ff
record_format dspace
spelling oai:doaj.org-article:063ce98188f84191bb67566055a282ff2021-11-19T00:05:48ZMobile Robot Localization Based on Gradient Propagation Particle Filter Network2169-353610.1109/ACCESS.2020.3031618https://doaj.org/article/063ce98188f84191bb67566055a282ff2020-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9226533/https://doaj.org/toc/2169-3536In order to solve the problem that the gradient information can’t propagate backward due to the non-differentiability of resampling process in the end-to-end training of Differentiable Particle Filters (DPFs) network model, a particle filter network with gradient propagation is proposed in this article. It uses a differentiable resampling process in the network and is called Gradient Propagation Particle Filter Network(GPPFN). Firstly, an auxiliary noise vector is added to approximate the particle distribution in the resampling process, so that it can be represented by a certain differentiable function. Then, the method is applied to GPPFN model to make the gradient information generated in the training process can be propagated. It can provide the information of previous prediction and update steps for the next step prediction, so as to improve the accuracy of robot localization. In this article, experiments are carried out in the simulation environment of Deepmind lab. GPPFN is used to train the motion model and measurement model respectively, and the training effect is expressed by likelihood degree. The lower the likelihood degree is, the more consistent the training results are. The experiments show that the likelihood loss of motion model and measurement model are reduced by using GPPFN. Finally, The GPPFN model is compared with other methods. The training results showed that compared with the others, the training results of GPPFN has lower error rate and better robustness.Heng ZhangJiemao WenYanli LiuWenqing LuoNaixue XiongIEEEarticleRobot localizationparticle filterdeep neural networkstate estimationElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENIEEE Access, Vol 8, Pp 188475-188487 (2020)
institution DOAJ
collection DOAJ
language EN
topic Robot localization
particle filter
deep neural network
state estimation
Electrical engineering. Electronics. Nuclear engineering
TK1-9971
spellingShingle Robot localization
particle filter
deep neural network
state estimation
Electrical engineering. Electronics. Nuclear engineering
TK1-9971
Heng Zhang
Jiemao Wen
Yanli Liu
Wenqing Luo
Naixue Xiong
Mobile Robot Localization Based on Gradient Propagation Particle Filter Network
description In order to solve the problem that the gradient information can’t propagate backward due to the non-differentiability of resampling process in the end-to-end training of Differentiable Particle Filters (DPFs) network model, a particle filter network with gradient propagation is proposed in this article. It uses a differentiable resampling process in the network and is called Gradient Propagation Particle Filter Network(GPPFN). Firstly, an auxiliary noise vector is added to approximate the particle distribution in the resampling process, so that it can be represented by a certain differentiable function. Then, the method is applied to GPPFN model to make the gradient information generated in the training process can be propagated. It can provide the information of previous prediction and update steps for the next step prediction, so as to improve the accuracy of robot localization. In this article, experiments are carried out in the simulation environment of Deepmind lab. GPPFN is used to train the motion model and measurement model respectively, and the training effect is expressed by likelihood degree. The lower the likelihood degree is, the more consistent the training results are. The experiments show that the likelihood loss of motion model and measurement model are reduced by using GPPFN. Finally, The GPPFN model is compared with other methods. The training results showed that compared with the others, the training results of GPPFN has lower error rate and better robustness.
format article
author Heng Zhang
Jiemao Wen
Yanli Liu
Wenqing Luo
Naixue Xiong
author_facet Heng Zhang
Jiemao Wen
Yanli Liu
Wenqing Luo
Naixue Xiong
author_sort Heng Zhang
title Mobile Robot Localization Based on Gradient Propagation Particle Filter Network
title_short Mobile Robot Localization Based on Gradient Propagation Particle Filter Network
title_full Mobile Robot Localization Based on Gradient Propagation Particle Filter Network
title_fullStr Mobile Robot Localization Based on Gradient Propagation Particle Filter Network
title_full_unstemmed Mobile Robot Localization Based on Gradient Propagation Particle Filter Network
title_sort mobile robot localization based on gradient propagation particle filter network
publisher IEEE
publishDate 2020
url https://doaj.org/article/063ce98188f84191bb67566055a282ff
work_keys_str_mv AT hengzhang mobilerobotlocalizationbasedongradientpropagationparticlefilternetwork
AT jiemaowen mobilerobotlocalizationbasedongradientpropagationparticlefilternetwork
AT yanliliu mobilerobotlocalizationbasedongradientpropagationparticlefilternetwork
AT wenqingluo mobilerobotlocalizationbasedongradientpropagationparticlefilternetwork
AT naixuexiong mobilerobotlocalizationbasedongradientpropagationparticlefilternetwork
_version_ 1718420648814968832