Search
Close this search box.

Robust 2D Assembly Sequencing via Geometric Planning with Learned Scores

Abstract

To compute robust 2D assembly plans, we present an approach that combines geometric planning with a deep neural network. We train the network using the Box2D physics simulator with added stochastic noise to yield robustness scores—the success probabilities of planned assembly motions. As running a simulation for every assembly motion is impractical, we train a convolutional neural network to map assembly operations, given as an image pair of the subassemblies before and after they are mated, to a robustness score. The neural network prediction is used within a planner to quickly prune out motions that are not robust. We demonstrate this approach on two-handed planar assemblies, where the motions are one-step translations. Results suggest that the neural network can learn robustness to plan robust sequences an order of magnitude faster than physics simulation.

(Left) A robust assembly operation, which we find as part of the assembly sequencing, versus an operation that is not (Right), which we opt to avoid

Links

Contacts

Tzvika Geft
Dan Halperin

@inproceedings{gtgh-rtasg-19,
author = {Tzvika Geft and Aviv Tamar and Ken Goldberg and Dan Halperin},
booktitle = {{IEEE} 15th International Conference on Automation Science and Engineering ({CASE})},
title = {Robust {2D} Assembly Sequencing via Geometric Planning with Learned Scores},
year = {2019},
pages = {1603–1610},
doi = {10.1109/COASE.2019.8843109}
}

@article{gtgh-rtasg-20,
author = {Tzvika Geft and Aviv Tamar and Ken Goldberg and Dan Halperin},
title = {Robust 2D Assembly Sequencing via Geometric Planning with Learned Scores},
journal = {CoRR},
volume = {abs/2009.09408},
year = {2020},
url = {https://arxiv.org/abs/2009.09408},
eprinttype = {arXiv},
eprint = {2009.09408},
timestamp = {Wed, 23 Sep 2020 15:51:46 +0200},
biburl = {https://dblp.org/rec/journals/corr/abs-2009-09408.bib},
bibsource = {dblp computer science bibliography, https://dblp.org}
}

Yair Oz - Webcreator

Contact

Skip to content