Model-based Behavioral Cloning with Future Image Similarity Learning

Alan Wu, AJ Piergiovanni, Michael S. Ryoo
Proceedings of the Conference on Robot Learning, PMLR 100:1062-1077, 2020.

Abstract

We present a visual imitation learning framework that enables learning of robot action policies solely based on expert samples without any robot trials. Robot exploration and on-policy trials in a real-world environment could often be expensive/dangerous. We present a new approach to address this problem by learning a future scene prediction model solely on a collection of expert trajectories consisting of unlabeled example videos and actions, and by enabling generalized action cloning using future image similarity. The robot learns to visually predict the consequences of taking an action, and obtains the policy by evaluating how similar the predicted future image is to an expert image. We develop a stochastic action-conditioned convolutional autoencoder, and present how we take advantage of future images for robot learning. We conduct experiments in simulated and real-life environments using a ground mobility robot with and without obstacles, and compare our models to multiple baseline methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v100-wu20b, title = {Model-based Behavioral Cloning with Future Image Similarity Learning}, author = {Wu, Alan and Piergiovanni, AJ and Ryoo, Michael S.}, booktitle = {Proceedings of the Conference on Robot Learning}, pages = {1062--1077}, year = {2020}, editor = {Kaelbling, Leslie Pack and Kragic, Danica and Sugiura, Komei}, volume = {100}, series = {Proceedings of Machine Learning Research}, month = {30 Oct--01 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v100/wu20b/wu20b.pdf}, url = {https://proceedings.mlr.press/v100/wu20b.html}, abstract = {We present a visual imitation learning framework that enables learning of robot action policies solely based on expert samples without any robot trials. Robot exploration and on-policy trials in a real-world environment could often be expensive/dangerous. We present a new approach to address this problem by learning a future scene prediction model solely on a collection of expert trajectories consisting of unlabeled example videos and actions, and by enabling generalized action cloning using future image similarity. The robot learns to visually predict the consequences of taking an action, and obtains the policy by evaluating how similar the predicted future image is to an expert image. We develop a stochastic action-conditioned convolutional autoencoder, and present how we take advantage of future images for robot learning. We conduct experiments in simulated and real-life environments using a ground mobility robot with and without obstacles, and compare our models to multiple baseline methods.} }
Endnote
%0 Conference Paper %T Model-based Behavioral Cloning with Future Image Similarity Learning %A Alan Wu %A AJ Piergiovanni %A Michael S. Ryoo %B Proceedings of the Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2020 %E Leslie Pack Kaelbling %E Danica Kragic %E Komei Sugiura %F pmlr-v100-wu20b %I PMLR %P 1062--1077 %U https://proceedings.mlr.press/v100/wu20b.html %V 100 %X We present a visual imitation learning framework that enables learning of robot action policies solely based on expert samples without any robot trials. Robot exploration and on-policy trials in a real-world environment could often be expensive/dangerous. We present a new approach to address this problem by learning a future scene prediction model solely on a collection of expert trajectories consisting of unlabeled example videos and actions, and by enabling generalized action cloning using future image similarity. The robot learns to visually predict the consequences of taking an action, and obtains the policy by evaluating how similar the predicted future image is to an expert image. We develop a stochastic action-conditioned convolutional autoencoder, and present how we take advantage of future images for robot learning. We conduct experiments in simulated and real-life environments using a ground mobility robot with and without obstacles, and compare our models to multiple baseline methods.
APA
Wu, A., Piergiovanni, A. & Ryoo, M.S.. (2020). Model-based Behavioral Cloning with Future Image Similarity Learning. Proceedings of the Conference on Robot Learning, in Proceedings of Machine Learning Research 100:1062-1077 Available from https://proceedings.mlr.press/v100/wu20b.html.

Related Material