Keywords: autonomous cooperation, multi-agent, genetic algorithm, evolutionary behaviour table system.

A collective robot system, which effectively works in such hazardous environments as fires, earthquakes, and outer space, is composed of autonomous and cooperation robots and is treated as a multi-agent problem. Generally, the behaviour rules of the multi-agent system are complicated in the design of intelligent programs because the behaviour of each agent is mutually influenced. Therefore, the efficiency of the intelligent system design technique must be enhanced in the intelligent collective robot system development.

To simplify the behaviour principle, actively using environmental information is effective as a methodology to install the autonomous behaviour principle in mobile robots. In other words, a mobile robot obtains environmental information from its sensors and acts based on the behaviour table that corresponds to the sensor information.

The principle can be discovered by the reflexion behaviour of such living things as social insects, and previous research exists [1,2]. Despite the small-scale system configuration, the robots of those researches realized such intelligent actions as the reflexion behaviour of insects. But problems exist, including the difficulty of establishing the behaviour principle for a specific purpose and the need to define the behaviour module.

On the other hand, some multi-robot research has addressed the above difficult problems in a single robot. Kube et al. [3,4] concentrated on the natural behaviour of social insects based on simple stimulus-response sequences and produced a collective robot system equipped with three IR sensors and two photo sensors. But in their research, the design of a stimulus-response rule base considerably increases the difficulty, because supervised learning requires another intelligent acquisition whenever additional sensors are mounted on the robot.

In the present research, we acquired the principle of autonomous behaviour in a behaviour-based robot and proposed an evolutionary behaviour table system (EBTS) using a genetic algorithm (GA) to acquire autonomy. Then we obtained a cooperative multi-agent system by a bottom-up method using an EBTS. The multi-agent system shows an advantage of the compound effect obtained by collectivity and multiple sensors. Furthermore, we acquired a principle of the concrete autonomous behaviour for a collective robot that solves the transportation problem as a collective task. In the future, the EBTS is expected to acquire complicated group intelligence extended by sensors and behaviour because the addition of more sensors and behaviour output are acceptable in the present system.

1

V. Braitenberg, "Adaptation in Natural and Artificial Systems", MIT Press, 1984.

2

R.A. Brooks, "A robust layered control system for a mobile robot", IEEE Journal of Robotics and Automation, RA-2(1), pp. 14-23, 1986.

3

C.R. Kube, H. Zhang, "Collective Robotic Intelligence", Second International Conference on Simulation of Adaptive Behavior, pp. 460-468, 1992.

4

C.R. Kube, H. Zhang, "Controlling Collective Tasks With An ALN", IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 289-293, 1993.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £95 +P&P)