Feature Learning for Scene Flow Estimation from LIDAR

Arash K. Ushani, Ryan M. Eustice
Proceedings of The 2nd Conference on Robot Learning, PMLR 87:283-292, 2018.

Abstract

To perform tasks in dynamic environments, many mobile robots must estimate the motion in the surrounding world. Recently, techniques have been developed to estimate scene flow directly from LIDAR scans, relying on hand-designed features. In this work, we build an encoding network to learn features from an occupancy grid. The network is trained so that these features are discriminative in finding matching or non-matching locations between successive timesteps. This learned feature space is then leveraged to estimate scene flow. We evaluate our method on the KITTI dataset and demonstrate performance that improves upon the accuracy of the current state-of-the-art. We provide an implementation of our method at https://github.com/aushani/flsf.

Cite this Paper

BibTeX
@InProceedings{pmlr-v87-ushani18a, title = {Feature Learning for Scene Flow Estimation from LIDAR}, author = {Ushani, Arash K. and Eustice, Ryan M.}, booktitle = {Proceedings of The 2nd Conference on Robot Learning}, pages = {283--292}, 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/ushani18a/ushani18a.pdf}, url = {https://proceedings.mlr.press/v87/ushani18a.html}, abstract = {To perform tasks in dynamic environments, many mobile robots must estimate the motion in the surrounding world. Recently, techniques have been developed to estimate scene flow directly from LIDAR scans, relying on hand-designed features. In this work, we build an encoding network to learn features from an occupancy grid. The network is trained so that these features are discriminative in finding matching or non-matching locations between successive timesteps. This learned feature space is then leveraged to estimate scene flow. We evaluate our method on the KITTI dataset and demonstrate performance that improves upon the accuracy of the current state-of-the-art. We provide an implementation of our method at https://github.com/aushani/flsf. } }
Endnote
%0 Conference Paper %T Feature Learning for Scene Flow Estimation from LIDAR %A Arash K. Ushani %A Ryan M. Eustice %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-ushani18a %I PMLR %P 283--292 %U https://proceedings.mlr.press/v87/ushani18a.html %V 87 %X To perform tasks in dynamic environments, many mobile robots must estimate the motion in the surrounding world. Recently, techniques have been developed to estimate scene flow directly from LIDAR scans, relying on hand-designed features. In this work, we build an encoding network to learn features from an occupancy grid. The network is trained so that these features are discriminative in finding matching or non-matching locations between successive timesteps. This learned feature space is then leveraged to estimate scene flow. We evaluate our method on the KITTI dataset and demonstrate performance that improves upon the accuracy of the current state-of-the-art. We provide an implementation of our method at https://github.com/aushani/flsf.
APA
Ushani, A.K. & Eustice, R.M.. (2018). Feature Learning for Scene Flow Estimation from LIDAR. Proceedings of The 2nd Conference on Robot Learning, in Proceedings of Machine Learning Research 87:283-292 Available from https://proceedings.mlr.press/v87/ushani18a.html.

Related Material