Intention-Net: Integrating Planning and Deep Learning for Goal-Directed Autonomous Navigation

Wei Gao, David Hsu, Wee Sun Lee, Shengmei Shen, Karthikk Subramanian
Proceedings of the 1st Annual Conference on Robot Learning, PMLR 78:185-194, 2017.

Abstract

How can a delivery robot navigate reliably to a destination in a new office building, with minimal prior information? To tackle this challenge, this paper introduces a two-level hierarchical approach, which integrates model-free deep learning and model-based path planning. At the low level, a neural-network motion controller, called the intention-net, is trained end-to-end to provide robust local navigation. The intention-net maps images from a single monocular camera and “intentions” directly to robot controls. At the high level, a path planner uses a crude map, e.g., a 2-D floor plan, to compute a path from the robot’s current location to the goal. The planned path provides intentions to the intention-net. Preliminary experiments suggest that the learned motion controller is robust against perceptual uncertainty and by integrating with a path planner, it generalizes effectively to new environments and goals.

Cite this Paper

BibTeX
@InProceedings{pmlr-v78-gao17a, title = {Intention-Net: Integrating Planning and Deep Learning for Goal-Directed Autonomous Navigation}, author = {Gao, Wei and Hsu, David and Lee, Wee Sun and Shen, Shengmei and Subramanian, Karthikk}, booktitle = {Proceedings of the 1st Annual Conference on Robot Learning}, pages = {185--194}, year = {2017}, editor = {Levine, Sergey and Vanhoucke, Vincent and Goldberg, Ken}, volume = {78}, series = {Proceedings of Machine Learning Research}, month = {13--15 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v78/gao17a/gao17a.pdf}, url = {https://proceedings.mlr.press/v78/gao17a.html}, abstract = {How can a delivery robot navigate reliably to a destination in a new office building, with minimal prior information? To tackle this challenge, this paper introduces a two-level hierarchical approach, which integrates model-free deep learning and model-based path planning. At the low level, a neural-network motion controller, called the intention-net, is trained end-to-end to provide robust local navigation. The intention-net maps images from a single monocular camera and “intentions” directly to robot controls. At the high level, a path planner uses a crude map, e.g., a 2-D floor plan, to compute a path from the robot’s current location to the goal. The planned path provides intentions to the intention-net. Preliminary experiments suggest that the learned motion controller is robust against perceptual uncertainty and by integrating with a path planner, it generalizes effectively to new environments and goals.} }
Endnote
%0 Conference Paper %T Intention-Net: Integrating Planning and Deep Learning for Goal-Directed Autonomous Navigation %A Wei Gao %A David Hsu %A Wee Sun Lee %A Shengmei Shen %A Karthikk Subramanian %B Proceedings of the 1st Annual Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2017 %E Sergey Levine %E Vincent Vanhoucke %E Ken Goldberg %F pmlr-v78-gao17a %I PMLR %P 185--194 %U https://proceedings.mlr.press/v78/gao17a.html %V 78 %X How can a delivery robot navigate reliably to a destination in a new office building, with minimal prior information? To tackle this challenge, this paper introduces a two-level hierarchical approach, which integrates model-free deep learning and model-based path planning. At the low level, a neural-network motion controller, called the intention-net, is trained end-to-end to provide robust local navigation. The intention-net maps images from a single monocular camera and “intentions” directly to robot controls. At the high level, a path planner uses a crude map, e.g., a 2-D floor plan, to compute a path from the robot’s current location to the goal. The planned path provides intentions to the intention-net. Preliminary experiments suggest that the learned motion controller is robust against perceptual uncertainty and by integrating with a path planner, it generalizes effectively to new environments and goals.
APA
Gao, W., Hsu, D., Lee, W.S., Shen, S. & Subramanian, K.. (2017). Intention-Net: Integrating Planning and Deep Learning for Goal-Directed Autonomous Navigation. Proceedings of the 1st Annual Conference on Robot Learning, in Proceedings of Machine Learning Research 78:185-194 Available from https://proceedings.mlr.press/v78/gao17a.html.

Related Material