Computational Models for Multiagent Coordination Analysis: Extending Distributed POMDP Models

Recently researchers in multiagent systems have begun to focus on formal POMDP (Partially Observable Markov Decision Process) models for analysis of multiagent coordination. However, prior work has mostly focused on analysis of communication, such as via the COM-MTDP (Communicative Markov Team Decision Problem) model. This paper provides two extensions to this prior work that goes beyond communication and analyzes other aspects of multiagent coordination. In particular, we first present a formal model called R-COM-MTDP that extends COM-MTDP to analyze team formation and reorganization algorithms. R-COM-MTDP enables a rigorous and systematic analysis of complexity-optimality tradeoffs in team (re)formation approaches in different domain types. It provides the worst-case complexity analysis of the team (re)formation under varying conditions, and illustrates under which conditions role decomposition can provide significant reductions in computational complexity. Next, we propose COM-MTDP as a formal framework to analyze DCSP (Distributed Constraint Satisfaction Problem) strategies for conflict resolution. Different DCSP strategies are mapped onto policies in the COM-MTDP model, and agents compare strategies by evaluating their mapped policies. Thus, the two COM-MTDP based methods could open the door to a range of novel analyses of multiagent team (re)formation, and facilitate automated selection of the most efficient strategy for a given situation.

[1]  Milind Tambe,et al.  Argumentation as distributed constraint satisfaction: applications and results , 2001, AGENTS '01.

[2]  Drew McDermott,et al.  The 1998 AI Planning Systems Competition , 2000, AI Mag..

[3]  Milind Tambe,et al.  Multiagent teamwork: analyzing the optimality and complexity of key theories and models , 2002, AAMAS '02.

[4]  Milind Tambe,et al.  Team Formation for Reformation in Multiagent Domains Like RoboCupRescue , 2002, RoboCup.

[5]  Steven Minton,et al.  Solving Large-Scale Constraint-Satisfaction and Scheduling Problems Using a Heuristic Repair Method , 1990, AAAI.

[6]  Makoto Yokoo,et al.  Asynchronous Weak-commitment Search for Solving Distributed Constraint Satisfaction Problems , 1995, CP.

[7]  Hiroaki Kitano,et al.  The RoboCup Synthetic Agent Challenge 97 , 1997, IJCAI.

[8]  Makoto Yokoo,et al.  Distributed constraint satisfaction algorithm for complex local problems , 1998, Proceedings International Conference on Multi Agent Systems (Cat. No.98EX160).

[9]  Michael P. Wellman,et al.  The 2001 trading agent competition , 2002, Electron. Mark..

[10]  Kee-Eung Kim,et al.  Learning to Cooperate via Policy Search , 2000, UAI.

[11]  Hiroaki Kitano,et al.  RoboCup Rescue: search and rescue in large-scale disasters as a domain for autonomous agents research , 1999, IEEE SMC'99 Conference Proceedings. 1999 IEEE International Conference on Systems, Man, and Cybernetics (Cat. No.99CH37028).

[12]  Wei-Min Shen,et al.  A Dynamic Distributed Constraint Satisfaction Approach to Resource Allocation , 2001, CP.

[13]  T. Yoshikawa.  Decomposition of dynamic team decision problems , 1978 .

[14]  Milind Tambe,et al.  Towards Flexible Teamwork , 1997, J. Artif. Intell. Res..