Teaching UAVs to Race With Observational Imitation Learning

Handle URI:
http://hdl.handle.net/10754/627342
Title:
Teaching UAVs to Race With Observational Imitation Learning
Authors:
Li, Guohao; Mueller, Matthias; Casser, Vincent; Smith, Neil; Michels, Dominik L.; Ghanem, Bernard ( 0000-0002-5534-587X )
Abstract:
Recent work has tackled the problem of autonomous navigation by imitating a teacher and learning an end-to-end policy, which directly predicts controls from raw images. However, these approaches tend to be sensitive to mistakes by the teacher and do not scale well to other environments or vehicles. To this end, we propose a modular network architecture that decouples perception from control, and is trained using Observational Imitation Learning (OIL), a novel imitation learning variant that supports online training and automatic selection of optimal behavior from observing multiple teachers. We apply our proposed methodology to the challenging problem of unmanned aerial vehicle (UAV) racing. We develop a simulator that enables the generation of large amounts of synthetic training data (both UAV captured images and its controls) and also allows for online learning and evaluation. We train a perception network to predict waypoints from raw image data and a control network to predict UAV controls from these waypoints using OIL. Our modular network is able to autonomously fly a UAV through challenging race tracks at high speeds. Extensive experiments demonstrate that our trained network outperforms its teachers, end-to-end baselines, and even human pilots in simulation. The supplementary video can be viewed at https://youtu.be/PeTXSoriflc
KAUST Department:
Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division; Electrical Engineering Program; Visual Computing Center (VCC)
Publisher:
arXiv
Issue Date:
3-Mar-2018
ARXIV:
arXiv:1803.01129
Type:
Preprint
Additional Links:
http://arxiv.org/abs/1803.01129v1; http://arxiv.org/pdf/1803.01129v1
Appears in Collections:
Other/General Submission; Electrical Engineering Program; Visual Computing Center (VCC); Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division

Full metadata record

DC FieldValue Language
dc.contributor.authorLi, Guohaoen
dc.contributor.authorMueller, Matthiasen
dc.contributor.authorCasser, Vincenten
dc.contributor.authorSmith, Neilen
dc.contributor.authorMichels, Dominik L.en
dc.contributor.authorGhanem, Bernarden
dc.date.accessioned2018-03-15T11:35:54Z-
dc.date.available2018-03-15T11:35:54Z-
dc.date.issued2018-03-03en
dc.identifier.urihttp://hdl.handle.net/10754/627342-
dc.description.abstractRecent work has tackled the problem of autonomous navigation by imitating a teacher and learning an end-to-end policy, which directly predicts controls from raw images. However, these approaches tend to be sensitive to mistakes by the teacher and do not scale well to other environments or vehicles. To this end, we propose a modular network architecture that decouples perception from control, and is trained using Observational Imitation Learning (OIL), a novel imitation learning variant that supports online training and automatic selection of optimal behavior from observing multiple teachers. We apply our proposed methodology to the challenging problem of unmanned aerial vehicle (UAV) racing. We develop a simulator that enables the generation of large amounts of synthetic training data (both UAV captured images and its controls) and also allows for online learning and evaluation. We train a perception network to predict waypoints from raw image data and a control network to predict UAV controls from these waypoints using OIL. Our modular network is able to autonomously fly a UAV through challenging race tracks at high speeds. Extensive experiments demonstrate that our trained network outperforms its teachers, end-to-end baselines, and even human pilots in simulation. The supplementary video can be viewed at https://youtu.be/PeTXSoriflcen
dc.publisherarXiven
dc.relation.urlhttp://arxiv.org/abs/1803.01129v1en
dc.relation.urlhttp://arxiv.org/pdf/1803.01129v1en
dc.rightsArchived with thanks to arXiven
dc.titleTeaching UAVs to Race With Observational Imitation Learningen
dc.typePreprinten
dc.contributor.departmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Divisionen
dc.contributor.departmentElectrical Engineering Programen
dc.contributor.departmentVisual Computing Center (VCC)en
dc.eprint.versionPre-printen
dc.identifier.arxividarXiv:1803.01129en
kaust.authorLi, Guohaoen
kaust.authorMueller, Matthiasen
kaust.authorCasser, Vincenten
kaust.authorSmith, Neilen
kaust.authorMichels, Dominik L.en
kaust.authorGhanem, Bernarden
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.