Refinement Acting vs. Simple Execution Guided by Hierarchical Planning

Simple item page
dc.contributor.advisorNau, Danaen_US
dc.contributor.authorBansod, Yashen_US
dc.contributor.departmentSystems Engineeringen_US
dc.contributor.publisherDigital Repository at the University of Marylanden_US
dc.contributor.publisherUniversity of Maryland (College Park, Md.)en_US
dc.date.accessioned2021-07-14T05:36:30Z
dc.date.available2021-07-14T05:36:30Z
dc.date.issued2021en_US
dc.description.abstractHumans have always reasoned about complex problems by organizing them into hierarchical structures. One approach to artificial intelligence planning is to design intelligent agents capable of breaking complex problems into multiple levels of abstraction so that at any one level, the problem becomes small and simple. However, for an agent to reason at multiple levels of abstraction, it needs knowledge at those abstraction levels. Hierarchical Task Network (HTN) planning allows us to do precisely that. This thesis presents a novel HTN planning algorithm that uses iterative tree traversal to refine HTNs. We also develop a purely reactive HTN acting algorithm using a similar procedure. Preserving the hierarchy in HTN plans can be helpful during execution. We make use of this fact to develop an algorithm for integrated HTN planning and acting. We show through experiments that our algorithm is an improvement over a widely used approach to planning and control.en_US
dc.identifierhttps://doi.org/10.13016/n4yb-kxqi
dc.identifier.urihttp://hdl.handle.net/1903/27468
dc.language.isoenen_US
dc.subject.pqcontrolledArtificial intelligenceen_US
dc.subject.pqcontrolledRoboticsen_US
dc.subject.pqcontrolledComputer scienceen_US
dc.subject.pquncontrolledArtificial Intelligence Actingen_US
dc.subject.pquncontrolledArtificial Intelligence Planningen_US
dc.subject.pquncontrolledHierarchical Planningen_US
dc.subject.pquncontrolledHierarchical Task Networksen_US
dc.subject.pquncontrolledIntegrated Planning and Actingen_US
dc.subject.pquncontrolledTask Refinementen_US
dc.titleRefinement Acting vs. Simple Execution Guided by Hierarchical Planningen_US
dc.typeThesisen_US

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Bansod_umd_0117N_21618.pdf
Size:
1.04 MB
Format:
Adobe Portable Document Format
Download

Collections

UMD Theses and Dissertations
Computer Science Theses and Dissertations