Please use this identifier to cite or link to this item: https://scholarbank.nus.edu.sg/handle/10635/212722
DC FieldValue
dc.titleDIFFERENTIABLE ROBOTICS: COMPOSITIONAL DEEP LEARNING WITH DIFFERENTIABLE ALGORITHM NETWORKS
dc.contributor.authorPETER KARKUS
dc.date.accessioned2021-12-31T18:01:31Z
dc.date.available2021-12-31T18:01:31Z
dc.date.issued2021-08-01
dc.identifier.citationPETER KARKUS (2021-08-01). DIFFERENTIABLE ROBOTICS: COMPOSITIONAL DEEP LEARNING WITH DIFFERENTIABLE ALGORITHM NETWORKS. ScholarBank@NUS Repository.
dc.identifier.urihttps://scholarbank.nus.edu.sg/handle/10635/212722
dc.description.abstractTowards human-level robot intelligence a central question is the architecture of robot learning: how is the system represented, and how is it trained? This thesis introduces the Differentiable Algorithm Network (DAN), a compositional architecture for designing robot learning systems. The DAN is composed of neural network modules, each encoding an algorithm and associated models; and it is trained end-to-end from data. The key idea is to make model-based algorithms differentiable and encode them in a generalized neural network, thus combining the model-based modular system design with data-driven end-to-end learning. Algorithms act as structural assumptions to moderate data requirements; end-to-end learning allows modules to adapt and compensate for imperfections. We introduce DANs for a range of domains: particle filter networks for visual localization; differentiable SLAM networks for visual SLAM; QMDP networks for partially observable planning. We also combine DAN modules for visual navigation in simulation and with a real-world quadruped robot.
dc.language.isoen
dc.subjectdeep learning, planning, model-based, model-free, end-to-end, modular, neural network
dc.typeThesis
dc.contributor.departmentINTEGRATIVE SCIENCES & ENGINEERING PROG
dc.contributor.supervisorHsu Xu Ye, David
dc.description.degreePh.D
dc.description.degreeconferredDOCTOR OF PHILOSOPHY (NUSGS)
dc.identifier.orcid0000-0002-1474-9771
Appears in Collections:Ph.D Theses (Open)

Show simple item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
KarkusP.pdf7.66 MBAdobe PDF

OPEN

NoneView/Download

Page view(s)

74
checked on Dec 1, 2022

Download(s)

29
checked on Dec 1, 2022

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.