Sim-to-Real Transfer for Vision-and-Language Navigation

Peter Anderson, Ayush Shrivastava, Joanne Truong, Arjun Majumdar, Devi Parikh, Dhruv Batra, Stefan Lee
Proceedings of the 2020 Conference on Robot Learning, PMLR 155:671-681, 2021.

Abstract

We study the challenging problem of releasing a robot in a previously unseen environment, and having it follow unconstrained natural language navigation instructions. Recent work on the task of Vision-and-Language Navigation (VLN) has achieved significant progress in simulation. To assess the implications of this work for robotics, we transfer a VLN agent trained in simulation to a physical robot. To bridge the gap between the high-level discrete action space learned by the VLN agent, and the robot’s low-level continuous action space, we propose a subgoal model to identify nearby waypoints, and use domain randomization to mitigate visual domain differences. For accurate sim and real comparisons in parallel environments, we annotate a 325m2 office space with 1.3km of navigation instructions, and create a digitized replica in simulation. We find that sim-to-real transfer to an environment not seen in training is successful if an occupancy map and navigation graph can be collected and annotated in advance (success rate of 46.8% vs. 55.9% in sim), but much more challenging in the hardest setting with no prior mapping at all (success rate of 22.5%).

Cite this Paper

BibTeX
@InProceedings{pmlr-v155-anderson21a, title = {Sim-to-Real Transfer for Vision-and-Language Navigation}, author = {Anderson, Peter and Shrivastava, Ayush and Truong, Joanne and Majumdar, Arjun and Parikh, Devi and Batra, Dhruv and Lee, Stefan}, booktitle = {Proceedings of the 2020 Conference on Robot Learning}, pages = {671--681}, year = {2021}, editor = {Kober, Jens and Ramos, Fabio and Tomlin, Claire}, volume = {155}, series = {Proceedings of Machine Learning Research}, month = {16--18 Nov}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v155/anderson21a/anderson21a.pdf}, url = {https://proceedings.mlr.press/v155/anderson21a.html}, abstract = {We study the challenging problem of releasing a robot in a previously unseen environment, and having it follow unconstrained natural language navigation instructions. Recent work on the task of Vision-and-Language Navigation (VLN) has achieved significant progress in simulation. To assess the implications of this work for robotics, we transfer a VLN agent trained in simulation to a physical robot. To bridge the gap between the high-level discrete action space learned by the VLN agent, and the robot’s low-level continuous action space, we propose a subgoal model to identify nearby waypoints, and use domain randomization to mitigate visual domain differences. For accurate sim and real comparisons in parallel environments, we annotate a 325m2 office space with 1.3km of navigation instructions, and create a digitized replica in simulation. We find that sim-to-real transfer to an environment not seen in training is successful if an occupancy map and navigation graph can be collected and annotated in advance (success rate of 46.8% vs. 55.9% in sim), but much more challenging in the hardest setting with no prior mapping at all (success rate of 22.5%).} }
Endnote
%0 Conference Paper %T Sim-to-Real Transfer for Vision-and-Language Navigation %A Peter Anderson %A Ayush Shrivastava %A Joanne Truong %A Arjun Majumdar %A Devi Parikh %A Dhruv Batra %A Stefan Lee %B Proceedings of the 2020 Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2021 %E Jens Kober %E Fabio Ramos %E Claire Tomlin %F pmlr-v155-anderson21a %I PMLR %P 671--681 %U https://proceedings.mlr.press/v155/anderson21a.html %V 155 %X We study the challenging problem of releasing a robot in a previously unseen environment, and having it follow unconstrained natural language navigation instructions. Recent work on the task of Vision-and-Language Navigation (VLN) has achieved significant progress in simulation. To assess the implications of this work for robotics, we transfer a VLN agent trained in simulation to a physical robot. To bridge the gap between the high-level discrete action space learned by the VLN agent, and the robot’s low-level continuous action space, we propose a subgoal model to identify nearby waypoints, and use domain randomization to mitigate visual domain differences. For accurate sim and real comparisons in parallel environments, we annotate a 325m2 office space with 1.3km of navigation instructions, and create a digitized replica in simulation. We find that sim-to-real transfer to an environment not seen in training is successful if an occupancy map and navigation graph can be collected and annotated in advance (success rate of 46.8% vs. 55.9% in sim), but much more challenging in the hardest setting with no prior mapping at all (success rate of 22.5%).
APA
Anderson, P., Shrivastava, A., Truong, J., Majumdar, A., Parikh, D., Batra, D. & Lee, S.. (2021). Sim-to-Real Transfer for Vision-and-Language Navigation. Proceedings of the 2020 Conference on Robot Learning, in Proceedings of Machine Learning Research 155:671-681 Available from https://proceedings.mlr.press/v155/anderson21a.html.

Related Material