Environmental robustness in multi-agent teams

Evolution has proven to be an effective method of training heterogeneous multi-agent teams of autonomous agents to explore unknown environments. Autonomous, heterogeneous agents are able to go places where humans are unable to go and perform tasks that would be otherwise dangerous or impossible to complete. However, a serious problem for practical applications of multi-agent teams is how to move from training environments to real-world environments. In particular, if the training environment cannot be made identical to the real-world environment how much will performance suffer? In this research we investigate how differences in training and testing environments affect performance. We find that while in general performance degrades from training to testing, for difficult training environments performance improves in the test environment. Further, we find distinct differences between the performance of different training algorithms with Orthogonal Evolution of Teams (OET) producing the best overall performance.

[1]  Terence Soule,et al.  Behavioral Diversity and a Probabilistically Optimal GP Ensemble , 2004, Genetic Programming and Evolvable Machines.

[2]  Sandip Sen,et al.  Evolving a Team , 1995 .

[3]  Tim Hendtlass,et al.  Collective intelligence and bush fire spotting , 2008, GECCO '08.

[4]  John R. Koza,et al.  Genetic programming - on the programming of computers by means of natural selection , 1993, Complex adaptive systems.

[5]  Robert Feldt,et al.  Generating multiple diverse software versions with genetic programming , 1998, Proceedings. 24th EUROMICRO Conference (Cat. No.98EX204).

[6]  Sean Luke,et al.  Cooperative Multi-Agent Learning: The State of the Art , 2005, Autonomous Agents and Multi-Agent Systems.

[7]  Terence Soule,et al.  Training Time and Team Composition Robustness in Evolved Multi-agent Systems , 2008, EuroGP.

[8]  James A. Foster,et al.  Fault-tolerant computing with N-version genetic programming , 2001 .

[9]  Ferat Sahin,et al.  Application of artificial immune system based intelligent multi agent model to a mine detection problem , 2002, IEEE International Conference on Systems, Man and Cybernetics.

[10]  Terence Soule,et al.  Improving Performance and Cooperation in Multi-Agent Systems , 2008 .

[11]  Lee Spector,et al.  Evolving teamwork and coordination with genetic programming , 1996 .

[12]  Xin Yao,et al.  Evolutionary ensembles with negative correlation learning , 2000, IEEE Trans. Evol. Comput..

[13]  Andrew B. Kahng,et al.  Cooperative Mobile Robotics: Antecedents and Directions , 1997, Auton. Robots.

[14]  David W. Opitz,et al.  Hazard assessment modeling: an evolutionary ensemble approach , 1999 .

[15]  Chandrika Kamath,et al.  Inducing oblique decision trees with evolutionary algorithms , 2003, IEEE Trans. Evol. Comput..

[16]  Terence Soule.  Cooperative Evolution on the Intertwined Spirals Problem , 2003, EuroGP.

[17]  James A. Foster,et al.  N-version genetic programming: a probabilistically optimal ensemble approach , 2002 .

[18]  Butong Zhang,et al.  Enhancing Robustness of Genetic Programming at the Species Level , 1996 .

[19]  T. Soule,et al.  Orthogonal Evolution of Teams: A Class of Algorithms for Evolving Teams with Inversely Correlated Errors , 2007 .

[20]  H. Iba.  Bagging, Boosting, and bloating in Genetic Programming , 1999 .

[21]  Terence Soule,et al.  Voting teams: a cooperative approach to non-typical problems using genetic programming , 1999 .

[22]  Wolfgang Banzhaf,et al.  Evolving Teams of Predictors with Linear Genetic Programming , 2001, Genetic Programming and Evolvable Machines.

[23]  Philippe Collard,et al.  Teams of Genetic Predictors for Inverse Problem Solving , 2005, EuroGP.