An Approach For Multi-Robot System Based on Agent Layered Architecture
Many agent-based architectures have been developed to facilitate the improvement of multi-robot systems which are able of performing robust cooperative work. This paper presents a layered architecture to simulate a robot-like agent. Different layers are defined for the software architecture with focusing on openness, flexibility, robustness and optimization properties. The proposed architecture aims to provide a well-structured and managed system for task execution, behaviour and decision-making of multi-robot system (MRS). Based on these ideas, we present a three-tiered agent architecture to harness many aspects of intelligence. Agents continuously observe the environment, in the reactive layer and act accordingly. The sequencing layer provides the means for planning, and task execution. We propose a fuzzy membership function, in the deliberate layer, to choose the appropriate plan during action selection using the satisfaction degree. This method is undertaken to illustrate, evaluate and find the most satisfactory plan alternative during the decision-making mechanism. We illustrate our line of thoughts with a Benchmark Production System used as a running example to explain our contribution.
Keywords- Agent-based architecture, Reactive layer, Sequencing layer, Deliberate layer, Satisfaction degree, Multi-robot system.