Analysis of Decision-making Architectures for a Self-assembling Reconfigurable Multi-agent Robot System: Annotated Bibliography

Tamara Blain

U.C. Berkeley

H. Bojinov, A. Casal, and T. Hogg. Multiagent control of self-reconfigurable robots. presented at the Int. Conf. on Multiagent Systems, 2000.

This paper details a control structure for the autonomous self-assembly of Proteo, a multi-agent robot, each agent shaped like a three dimensionalrhombic dodecahedron (3D analog of the hexagon). Seeds, which are modules that attract other modules, emit 'scents' to attract modules to themselves. Control programs in the modules are not directly notified of the outcome of their movement requests implying a module could get stuck trying the same move repeatedly. But module decisions are affected by the behavior of other modules, thus over time the stuck module would decide on a new behavior.

R. Peter Bonasso and David Kortenkamp. Using a layered control architecture to alleviate planning with incomplete information. In Proceedings of the AAAI Spring Symposium \Planning with Incomplete Information for Robot Problems, pages 1{4, Menlo Park, CA, March 1996. AAAI Press.

The authors discuss a 3-tiered architecture consisting of planning, sequencing and control for individual goal-directed robots in natural environments. The authors propose reaction action packages (RAPs), which are programmed procedures for accomplishing tasks, for decision making. The idea is that a goal could be broken down into RAPs, which in turn could be used to select specific skills in an agent's skill set.

Y. Uny Cao, Alex Fukunaga, Andrew Kahng, and Frank Meng. Cooperative mobile robotics: Antecedents and directions. In Proceedings of 1995 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS ’95) , pages 226-234, 1995.

This work is an overview of key theoretical bases for research in the field of cooperative mobile robotics. Discussions of distributed problem solving, distributed artificial intelligence, and examplary distributed robot architectures are included.

R. Chandrasekar. Information theoretic self-organization of multiple agents. Masters thesis, U. of Florida, Summer 2003.

This comprehensive work discusses swarms, multiagent robot systems, and their various control strategies. The author introduces the Information Particle Interaction Method (IPIM), based on particle potential fields, and proposes a model for the self-organization of multi-agent robots through pair-wise interactions of 'particles'.

Dudeck, G., Jenkin, M., Milios, E. & Wilkes, D. A taxonomy for multi-agent robotics. Autonomous Robots, 3, 375 ± 397, 1996.

This paper proposes a common language for the description and evaluation of multiagent systems.

J.R.Kok, M. T. J. Spaan, and N. Vlassis. Multi-robot decision making using coordination graphs. In Proceedings of the 11th International Conference on Advanced Robotics, ICAR'03, pp. 1124–1129, Coimbra, Portugal, June 2003.

In this proposed coordination graph based algorithm, each node represents an agent, and an edge indicates that the corresponding agents have to coordinate their actions. Each agent is assigned a role which is computed based on a role assignment function. Agents also have individual payoff functions which map the selected joint action to a real value. Each agent selects an action from its action set, and then receives a payoff based on the actions selected by all agents. The goal of the agents is to select, via their individual decisions, the most profitable joint action. The algorithm assumes that each agent knows its neighbors in the graph, but not necessarily their payoff function which might depend on other agents.

A. Kamimura, H. Kurokawa, E. Yoshida, S. Murata, K. Tomita and S. Kokaji: Distributed Adaptive Locomotion by a Modular Robotic System, M-TRAN II (From Local Adaptation to Global Coodinated Motion using CPG Controllers). In Proceedings of 2004 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS2004), pp. 2370-2377, 2004.

This work concerns the M-TRAN II, a modular, mobile, multiagent robot system. A distributed control method for locomotion is introduced, which is based on a real-time central pattern generator (CPG) model. Hardware experiments testing the implementation of CPG-based control are detailed and discussed.

Y. Liu and K. M. Passino. Swarm Intelligence: Literature Overview. Dept.
of Electrical Engineering, Ohio State University, 2000.

The authors present an explanation of swarm intelligence and its relevance to robotics, and discuss the architecures of demonstrative multiagent robot systems.

T. Laengle, T. C. Lueth, U. Rembold, and H. Woern. A distributed control architecture for autonomous mobile robots – implementation of the Karlsruhe Multi- agent Robot Architecture (KAMARA)”, Advanced Robotics, Volume 12, No. 4, pp. 411-431, 1998.

The authors argue the case for distributed control of multi-agent robots. The arguments refer to the KAMRO robot, whose agents remain in a fixed configuration. It presents a decentralized/centralized architecture where 'special agents' each have temporary control over the bodies their team of agents for the completion of a task.

A. Martinoli. Swarm intelligence in autonomous collective robotics: From tools to the analysis and synthesis of distributed control strategies. PhD Thesis, Computer Science Department, Ecole Polytechnique F`ed`erale de Lausanne, 1999.

This comprehensive work focuses on the Khepera robot. Chapter 2 gives us a survey of the types of existing control architectures. Chapter 5 discusses the results of various experiments implementing select control architectures.

Mondada, F., Pettinaro, G.C., Kwee, I., Guignard, A., Gambardella, L.M., Floreano, D., Nolfi, S., Deneubourg, J.-L., Dorigo, M. SWARM-BOT: A Swarm of Autonomous Mobile Robots with Self-Assembling Capabilities. In Hemelrijk, C.K. (ed.): International Workshop on Self-Organisation and Evolution of Social Behaviour. Swiss Federal Institute of Technology, Zurich Switzerland (2002)

The SWARMBOT concept is first proposed and explained here.

F. Mondada, G.C. Pettinaro, A. Guignard, I.V. Kwee, D. Floreano, J.-L. Deneubourg, S. Nol, L. M. Gambardella, and M. Dorigo. SWARM-BOT: a new distributed robotic concept. Autonomous Robots. In Press.

The swarm-bot concept is fleshed out and is discussed in terms versatility, robustness, and ability to navigate rough terrain. The hardware and software details of its implementation and the swarmbot-3d simulator written for testing are discussed.

A. Oller, J.L. de la Rosa, R. Garcia, J.A. Ramon, and A. Figueras. Micro-robots playing soccer games: a real implementation based on a multiagent decision-making structure. International Journal of Intelligent Automation and Soft Computing, 6(1):65–74, 2000. Special Issue on Soccer Robotics: Micro-robot WorldCup Soccer Tournament’97.

The competive soccer playing robots described utilize a hierarchical centralized architecture. Path planning and the calculation of set points takes place at the supervisory level. Execution of set points is handled by the control level, and individual agent behaviour decisions at the agent level. The authors have written their own agent oriented language in order to explicitly represent intention.

Parks, D. Agent-Oriented Programming: A Practical Evaluation,

1998

This work is an overview of agent-oriented programming and a critique of 3 agent-oriented languages, Agent-0, Agent-k, and PLACA.

P. Pirjanian, T.L. Huntsberger, A. Trebi-Ollennu, H. Aghazarian, H. Das, S. Joshi, and P.S. Schenker. CAMPOUT: A control architecture for multi-robot planetary outposts. In Proc. SPIE Symposium on Sensor Fusion and Decentralized Control in Robotic Systems III, Vol. 4196, Boston, MA, Nov. 2000, pp. 221-230.

The authors favor a distributed architecture where active collaboration between agents, each with its own library of behaviors, results in behavior exchange. Thus group behavior consists of a a network of behaviors across a distributed group of robots. Behaviors are hierarchical perception to action mapping modules that, based on selective sensory information, produce (recommendations for) actions. The architecture proposed also includes learning capabilities in order for each agent and it's abilities to evolve and improve.

E. Sahin, T. H. Labella, V. Trianni, J.-L. Deneubourg, P. Rasse, D. Floreano, L. M. Gambardella, F. Mondada, S. Nolfi, and M. Dorigo. SWARMBOT: Pattern formation in a swarm of self-assembling mobile robots. In Proceedings of the IEEE International Conference on Systems, Man and Cybernetics, IEEE Press, Piscataway, NJ, 2002.

The authors detail the requirements and limitations of the SWARM-BOT and discuss their simulations of various benchmarks involving self-organization.

Trianni V., Nolfi S. & Dorigo M. Cooperative hole-avoidance in a swarm-bot. Robotics & Autonomous Systems, (in press).

This paper is a study of cooperative behavior in a swarm-bot, focusing on the performance of an evolutionary algorithm in the simulation of a hole avoidance task.

J. Wessnitzer, A. Adamatzky, and C. Melhuish. Towards Self-Organising Structure Formations: A Decentralized Approach. Proceedings of the 6th European Conference on Advances in Artificial Life, p.573-581, September 10-14, 2001.

The authors apply methods and ideas used in cellular automata theory and complex systems theory to a decision-making framework for self-assembling multiagent mobile robot systems. They analyze the framework by simulating various approaches to spatial self-organiation.

P. Will, A. Costano, and W. M. Shen. Robot modularity for self-reconfiguration. Proc. SPIE, vol. 3839, 236/245 (1999).

This work introduces the concept of 'hormones' that travel among agents triggering different agents into particular actions. The idea is that the hormone, when coupled with the agent's role in the system, can be used to index into a table of behaviours. The authors argue the case for hormone-based control of distributed robot systems.