Article Preview
Top1. Introduction
Nowadays, there are two main Robotic System applications: (1) artificial intelligence robotics which is concerned on how a robot can handle unpredictable events in an unstructured world, (2) industrial robotics which concentrates on robot's dynamics and kinematics. In our work, we study rather the artificial Intelligence robotics and we map agent technology on the world of robotics. This mapping seems at the first look surprising as a robot is considered a physical entity ensuring a variety of tasks. However, there are many similarities between Agent and Robot which pushes in this way (for example, the robot like the agent has several activities and responds to its environment). Besides, we believe that the multi-agent technology applied to robotic system is able to increase the flexibility of the system as a whole. This article is concerned with two important matters: how to define the Multi-Robot System (MRS) in a manner such that it has more utility to deploy it, and how to use such a MRS for the advanced software. The Multi-Robot System must discover the action to be taken by supervising the application and its environment and analyzing the data obtained.
With Multi-Robot System, we face two important matters:
- 1.
The detection of a need for action: the need for action must be discovered by supervising the application and its environment and analyzing data obtained.
- 2.
The planning of the action: it consists to envisage the action (by proposing which modifications need to be made) and by programming it.
Using a multi-agent approach, the robot’s architecture can be decomposed into flexible autonomous subsystems (agents). The architecture can then be described at a higher level, defining the agents that have to be in the system, the role of each of them, the interactions among them, the actions each of them performs, and the resources they need. Since the multi-agent system is inherently multi-threaded, each agent has its own thread of control; each agent decides whether or not to perform an action at the request of another agent (autonomy); agents establish agreements among themselves, while keeping their autonomy sharing their knowledge and acting together to accomplish specific common goals. Agents need to interact to coordinate their activities so that control of the robot is achieved. All of those processes, the agent’s own decision making, interaction and coordination need to be highlighted. To do so, we propose the design of a Robotic Agent according to the 5 Capabilities model (5C) proposed by (van Aart, 2004; van Aart et al., 2004).
The 5 Capabilities model is separated into five dimensions: Environment, Self, Planning, Competence and Communication. These dimensions are said models where each model represents one specific capability of the robotic agent. First of all, a robotic agent needs to interact with the environment in which it operates thanks to sensors (providing data) and actuators (executing actions) therefore we define the Environment Model. To know what tasks to be executed, we define the Self Model. The self model is used to know the robotic agent's perception of its own being and state. In other terms, it consists of ongoing tasks. The planning of ongoing tasks is the concern of the Planner Model. A planner model is ensuring some kind of reasoning about task selection, execution control, time monitoring and emergency handling. The Competence Model ensures the methods the abilities and the knowledge that enables the robotic agent to execute the task that is designed for it. The multi-robot system has the appropriate Communication Model in order to avoid non-feasible, unsecured and fortuitous actions that can provoke undesirable results by a single robot for the whole system, the different robotic agents have to interact together following a specific communication protocol.
This paper introduces a simple Benchmark Production System that will be used throughout this article to illustrate our contribution which is developed as Robot-based application. We implement the Benchmark Production System in a free platform which is JADE (JavaTM Agent Development) Framework. JADE is a platform to develop multi-agent systems in compliance with the FIPA specifications (Vitabile et al., 2009; Bordini et al., 2006).