Show simple item record

dc.contributor.authorLi, Guohao
dc.contributor.authorMüller, Matthias
dc.contributor.authorCasser, Vincent Michael
dc.contributor.authorSmith, Neil
dc.contributor.authorMichels, Dominik L.
dc.contributor.authorGhanem, Bernard
dc.date.accessioned2019-12-02T12:48:23Z
dc.date.available2018-03-15T11:35:54Z
dc.date.available2019-04-18T13:34:39Z
dc.date.available2019-12-02T12:48:23Z
dc.date.issued2019-06
dc.identifier.citationLi, G., Mueller, M., Michael Casser, V., Smith, N., Michels, D., & Ghanem, B. (2019). OIL: Observational Imitation Learning. Robotics: Science and Systems XV. doi:10.15607/rss.2019.xv.005
dc.identifier.doi10.15607/RSS.2019.XV.005
dc.identifier.urihttp://hdl.handle.net/10754/627342
dc.description.abstractRecent work has explored 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 Observational Imitation Learning (OIL), a novel imitation learning variant that supports online training and automatic selection of optimal behavior by observing multiple imperfect teachers. We apply our proposed methodology to the challenging problems of autonomous driving and UAV racing. For both tasks, we utilize the Sim4CV simulator that enables the generation of large amounts of synthetic training data and also allows for online learning and evaluation. We train a perception network to predict waypoints from raw image data and use OIL to train another network to predict controls from these waypoints. Extensive experiments demonstrate that our trained network outperforms its teachers, conventional imitation learning (IL) and reinforcement learning (RL) baselines and even humans in simulation.
dc.description.sponsorshipThis work was supported by the King Abdullah University of Science and Technology (KAUST) Office of Sponsored Research through the Visual Computing Center (VCC) funding.
dc.publisherarXiv
dc.relation.ispartofProject:https://sites.google.com/kaust.edu.sa/oil/
dc.relation.urlhttp://www.roboticsproceedings.org/rss15/p05.html
dc.rightsArchived with thanks to Proceedings of Robotics: Science and Systems
dc.titleOIL: Observational Imitation Learning
dc.typeConference Paper
dc.contributor.departmentElectrical Engineering Program
dc.contributor.departmentVisual Computing Center (VCC)
dc.contributor.departmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division
dc.contributor.departmentComputer Science Program
dc.conference.dateJune 22-26, 2019
dc.conference.nameRobotics: Science and Systems XV
dc.conference.locationFreiburg im Breisgau, Germany
dc.eprint.versionPublisher's Version/PDF
dc.relation.embedded<iframe width="560" height="315" src="https://www.youtube.com/embed/_rhq8a0qgeg" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
dc.identifier.arxividarXiv:1803.01129
kaust.personLi, Guohao
kaust.personMueller, Matthias
kaust.personCasser, Vincent Michael
kaust.personSmith, Neil
kaust.personMichels, Dominik L.
kaust.personGhanem, Bernard
dc.relation.issupplementedbyURL:https://youtu.be/_rhq8a0qgeg
refterms.dateFOA2018-06-14T05:50:42Z
display.relations<b>Is Part Of:</b> <br/> <ul> <li><i>[Project]</i> <br/> OIL: Observational Imitation Learning. URL: <a href="https://sites.google.com/kaust.edu.sa/oil/">https://sites.google.com/kaust.edu.sa/oil/</a></li></ul> <b>Is Supplemented By:</b> <br/><ul> <li><i>[Video]</i> <br/> OIL: Observational Imitation Learning. URL: <a href="https://youtu.be/_rhq8a0qgeg">https://youtu.be/_rhq8a0qgeg</a></li></ul>
kaust.acknowledged.supportUnitOffice of Sponsored Research
kaust.acknowledged.supportUnitVisual Computing Center (VCC)
dc.date.posted2018-03-03


Files in this item

Thumbnail
Name:
p05.pdf
Size:
3.517Mb
Format:
PDF
Description:
Conference Paper

This item appears in the following Collection(s)

Show simple item record

VersionItemEditorDateSummary

*Selected version