Utilizing task partitioning for self-organized allocation in multi-robot systems
General Material Designation
[Thesis]
First Statement of Responsibility
Nourhan M. Elsayed
Subsequent Statement of Responsibility
Al-Wahedi, Khaled
.PUBLICATION, DISTRIBUTION, ETC
Name of Publisher, Distributor, etc.
The Petroleum Institute (United Arab Emirates)
Date of Publication, Distribution, etc.
2015
PHYSICAL DESCRIPTION
Specific Material Designation and Extent of Item
131
GENERAL NOTES
Text of Note
Committee members: Al Durra, Ahmed; AlHammadi, Khalid; Jarrar, Firas Salah
NOTES PERTAINING TO PUBLICATION, DISTRIBUTION, ETC.
Text of Note
Place of publication: United States, Ann Arbor; ISBN=978-1-339-16114-3
DISSERTATION (THESIS) NOTE
Dissertation or thesis details and type of degree
M.S.
Discipline of degree
Electrical Engineering
Body granting the degree
The Petroleum Institute (United Arab Emirates)
Text preceding or following the note
2015
SUMMARY OR ABSTRACT
Text of Note
Despite the long-going research in multi-robot systems, a big gap of knowledge still exists in coordination mechanisms such as task allocation. The current literature provides neither a well-dened understanding of self-organized multi-robot task allo- cation problems nor tools for designing and evaluating such systems. In this work, a new method for self-organized task allocation utilizing task partitioning is proposed. A complex foraging problem with multiple sources and nests is broken down into two smaller sequential subtasks. The two subtasks describe how the transportation of multiple objects is handled by two heterogeneous robots. We rst describe the subtasks and study their properties to provide a systematic way of modeling the system. We then propose dierent algorithms to achieve these subtasks through self- organized task allocation. One of these algorithms utilize topological sorting which, till now, has not been applied to the eld of task allocation in robotics. Other algo- rithms developed utilize integer programming and genetic algorithms for an optimal task assignment. We compare and analyze the performance of these algorithms under dierent conditions. We validate the system using the Webots real-time simulator.