2026-09-19T07:41:20Zhttps://dspace.mit.edu/server/oai/request

oai:dspace.mit.edu:1721.1/1201952022-01-13T07:54:05Zcom_1721.1_7582com_1721.1_7581col_1721.1_131023

Jonathan P. How. Buckman, Noam (Noam M.) Massachusetts Institute of Technology. Department of Mechanical Engineering. Massachusetts Institute of Technology. Department of Mechanical Engineering 2019-02-05T15:17:44Z 2019-02-05T15:17:44Z 2018 2018 http://hdl.handle.net/1721.1/120195 1083126862 Thesis: S.M., Massachusetts Institute of Technology, Department of Mechanical Engineering, 2018. This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections. Cataloged from student-submitted PDF version of thesis. Includes bibliographical references (pages 103-110). In time-sensitive and dynamic missions, autonomous vehicles must respond quickly to new information and objectives. In the case of dynamic task allocation, a team of agents are presented with a new, unknown task that must be allocated with their original allocations. This is exacerbated further in decentralized settings where agents are limited to utilizing local information during the allocation process. This thesis presents a fully decentralized, dynamic task allocation algorithm that extends the Consensus-Based Bundle Algorithm (CBBA) to allow for allocating new tasks. Whereas static CBBA requires a full resetting of previous allocations, CBBA with Partial Replanning (CBBA-PR) enables the agents to only partially reset their allocations to efficiently and quickly allocate a new task. By varying the number of existing tasks that are reset during replan, the team can trade-off convergence speed with amount of coordination. By specifically choosing the lowest bid tasks for resetting, CBBA-PR is shown to converge linearly with the number of tasks reset and the network diameter of the team. In addition, limited replanning methods are presented for scenarios without sufficient replanning time. These include a single reset bidding procedure for agents at capacity, a no-replanning heuristic that can identify scenarios that does not require replanning, and a subteam formation algorithm for reducing the network diameter. Finally, this thesis describes hardware and simulation experiments used to explore the effects of ad-hoc, decentralized communication on consensus algorithms and to validate the performance of CBBA-PR. by Noam Buckman. S.M. 110 pages eng Massachusetts Institute of Technology MIT theses are protected by copyright. They may be viewed, downloaded, or printed from this source but further reproduction or distribution in any format is prohibited without written permission. http://dspace.mit.edu/handle/1721.1/7582 Mechanical Engineering. Decentralized task allocation for dynamic, time-sensitive tasks Thesis application/pdf false Publication unknown unknown <Publication xmlns="https://www.openaire.eu/cerif-profile/1.1/" id="03841de1-7a48-4d21-a8b7-1450fcbb43f6"> <Type xmlns="https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types">http://purl.org/coar/resource_type/c_1843</Type> <Language>eng</Language> <Title>Decentralized task allocation for dynamic, time-sensitive tasks</Title> <PublishedIn> <Publication> </Publication> </PublishedIn> <PublicationDate>2018</PublicationDate> <Authors> <Author> <DisplayName>Buckman, Noam (Noam M.)</DisplayName> <Affiliation> <OrgUnit> </OrgUnit> </Affiliation> </Author> </Authors> <Editors> </Editors> <Publishers> <Publisher> <DisplayName>Massachusetts Institute of Technology</DisplayName> <OrgUnit /> </Publisher> </Publishers> <License>http://dspace.mit.edu/handle/1721.1/7582</License> <Keyword>Mechanical Engineering.</Keyword> <Abstract>In time-sensitive and dynamic missions, autonomous vehicles must respond quickly to new information and objectives. In the case of dynamic task allocation, a team of agents are presented with a new, unknown task that must be allocated with their original allocations. This is exacerbated further in decentralized settings where agents are limited to utilizing local information during the allocation process. This thesis presents a fully decentralized, dynamic task allocation algorithm that extends the Consensus-Based Bundle Algorithm (CBBA) to allow for allocating new tasks. Whereas static CBBA requires a full resetting of previous allocations, CBBA with Partial Replanning (CBBA-PR) enables the agents to only partially reset their allocations to efficiently and quickly allocate a new task. By varying the number of existing tasks that are reset during replan, the team can trade-off convergence speed with amount of coordination. By specifically choosing the lowest bid tasks for resetting, CBBA-PR is shown to converge linearly with the number of tasks reset and the network diameter of the team. In addition, limited replanning methods are presented for scenarios without sufficient replanning time. These include a single reset bidding procedure for agents at capacity, a no-replanning heuristic that can identify scenarios that does not require replanning, and a subteam formation algorithm for reducing the network diameter. Finally, this thesis describes hardware and simulation experiments used to explore the effects of ad-hoc, decentralized communication on consensus algorithms and to validate the performance of CBBA-PR.</Abstract> <Access xmlns="http://purl.org/coar/access_right" > </Access> </Publication>