Type of Document	Dissertation
Author	Zhang, Kai
Author's Email Address	kz06@fsu.edu
URN	etd-07112010-134821
Title	Stochastic Clustering Auctions for Cooperative Task Allocation
Degree	Doctor of Philosophy
Department	Mechanical Engineering, Department of
Advisory Committee	Advisor Name Title Emmanuel G. Collins Committee Chair David Cartes Committee Member Leon van Dommelen Committee Member Adrian Barbu University Representative
Keywords	Auctions and Market-based Systems Optimal Task Allocation Distributed Robot Systems Networked Agents Markov Chain Monte Carlo Simulated Annealing Swendsen-Wang Method
Date of Defense	2010-06-28
Availability	unrestricted
Abstract This dissertation considers the problem of optimal task allocation for heterogeneous teams, e.g., teams of heterogeneous robots or human-robot teams. It is well known that this problem is NP hard and hence computationally feasible approaches must develop an approximate solution. It will be shown that the global cost of the task allocations obtained with fast greedy algorithms can be improved upon by using a class of cooperative auction methods called Stochastic Clustering Auctions (SCAs). SCAs use stochastic transfers or swaps between the task clusters assigned to each team member, allow both uphill and downhill cost movements, and rely on simulated annealing. The choice of a key annealing parameter and turning the uphill movements on and off enables the converged solution of a SCA to slide in the region between the global optimal performance and the performance associated with a random allocation. The first SCA developed in this research, called GSSCA, is based on a Gibbs sampler, constrains the stochastic cluster reallocations to simple single transfers or swaps, and is applicable to heterogeneous teams. For homogeneous teams this dissertation presents a new and more efficient SCA, called SWSCA, based on the generalized Swendsen-Wang method, which enables more complex and efficient movements between clusters by connecting tasks that appear to be synergistic and then stochastically reassigning these connected tasks. For heterogeneous teams this dissertation proposes a HYbrid Stochastic Clustering Auction, called HYSCA. In HYSCA the auctioneer makes stochastic movements with single tasks when the auctioneer negotiates with heterogeneous agents and makes stochastic movements with interconnected tasks when the auctioneer negotiates with homogeneous agents. For centralized auctioning extensive numerical experiments were used to compare GSSCA with greedy auctioning methods for homogeneous teams and heterogeneous teams in terms of costs and computational and communication requirements. A series of random simulations showed that SWSCA was able to obtain significantly greater cost improvements than GSSCA for both the greedy and non-greedy cases for homogeneous teams and HYSCA was able to obtain significantly greater cost improvements than GSSCA for heterogenous teams. For distributed auctioning simulation results are presented from random scenarios and for selected benchmark auction patterns with a focus on a comparison of the performance achieved with distributed and centralized GSSCA. Finally the distributed SWSCA and HYSCA is evaluated in numerical experiments in which the communication links between agents were motivated by a generic topology called a ``scale free network.'
Files	Filename       Size       Approximate Download Time (Hours:Minutes:Seconds)     28.8 Modem   56K Modem   ISDN (64 Kb)   ISDN (128 Kb)   Higher-speed Access    Zhang_K_Dissertation_2010.pdf 3.31 Mb 00:15:19 00:07:52 00:06:53 00:03:26 00:00:17