Particle Filter Networks with Application to Visual Localization

Peter Karkus, David Hsu, Wee Sun Lee
Proceedings of The 2nd Conference on Robot Learning, PMLR 87:169-178, 2018.

Abstract

Particle filtering is a powerful approach to sequential state estimation and finds application in many domains, including robot localization, object tracking, etc. To apply particle filtering in practice, a critical challenge is to construct probabilistic system models, especially for systems with complex dynamics or rich sensory inputs such as camera images. This paper introduces the Particle Filter Network (PFnet), which encodes both a system model and a particle filter algorithm in a single neural network. The PF-net is fully differentiable and trained end-to-end from data. Instead of learning a generic system model, it learns a model optimized for the particle filter algorithm. We apply the PF-net to a visual localization task, in which a robot must localize in a rich 3-D world, using only a schematic 2-D floor map. In simulation experiments, PF-net consistently outperforms alternative learning architectures, as well as a traditional model-based method, under a variety of sensor inputs. Further, PF-net generalizes well to new, unseen environments.

Cite this Paper

BibTeX
@InProceedings{pmlr-v87-karkus18a, title = {Particle Filter Networks with Application to Visual Localization}, author = {Karkus, Peter and Hsu, David and Lee, Wee Sun}, booktitle = {Proceedings of The 2nd Conference on Robot Learning}, pages = {169--178}, year = {2018}, editor = {Billard, Aude and Dragan, Anca and Peters, Jan and Morimoto, Jun}, volume = {87}, series = {Proceedings of Machine Learning Research}, month = {29--31 Oct}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v87/karkus18a/karkus18a.pdf}, url = {https://proceedings.mlr.press/v87/karkus18a.html}, abstract = {Particle filtering is a powerful approach to sequential state estimation and finds application in many domains, including robot localization, object tracking, etc. To apply particle filtering in practice, a critical challenge is to construct probabilistic system models, especially for systems with complex dynamics or rich sensory inputs such as camera images. This paper introduces the Particle Filter Network (PFnet), which encodes both a system model and a particle filter algorithm in a single neural network. The PF-net is fully differentiable and trained end-to-end from data. Instead of learning a generic system model, it learns a model optimized for the particle filter algorithm. We apply the PF-net to a visual localization task, in which a robot must localize in a rich 3-D world, using only a schematic 2-D floor map. In simulation experiments, PF-net consistently outperforms alternative learning architectures, as well as a traditional model-based method, under a variety of sensor inputs. Further, PF-net generalizes well to new, unseen environments. } }
Endnote
%0 Conference Paper %T Particle Filter Networks with Application to Visual Localization %A Peter Karkus %A David Hsu %A Wee Sun Lee %B Proceedings of The 2nd Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2018 %E Aude Billard %E Anca Dragan %E Jan Peters %E Jun Morimoto %F pmlr-v87-karkus18a %I PMLR %P 169--178 %U https://proceedings.mlr.press/v87/karkus18a.html %V 87 %X Particle filtering is a powerful approach to sequential state estimation and finds application in many domains, including robot localization, object tracking, etc. To apply particle filtering in practice, a critical challenge is to construct probabilistic system models, especially for systems with complex dynamics or rich sensory inputs such as camera images. This paper introduces the Particle Filter Network (PFnet), which encodes both a system model and a particle filter algorithm in a single neural network. The PF-net is fully differentiable and trained end-to-end from data. Instead of learning a generic system model, it learns a model optimized for the particle filter algorithm. We apply the PF-net to a visual localization task, in which a robot must localize in a rich 3-D world, using only a schematic 2-D floor map. In simulation experiments, PF-net consistently outperforms alternative learning architectures, as well as a traditional model-based method, under a variety of sensor inputs. Further, PF-net generalizes well to new, unseen environments.
APA
Karkus, P., Hsu, D. & Lee, W.S.. (2018). Particle Filter Networks with Application to Visual Localization. Proceedings of The 2nd Conference on Robot Learning, in Proceedings of Machine Learning Research 87:169-178 Available from https://proceedings.mlr.press/v87/karkus18a.html.

Related Material