Diseño de reconfiguración dinámica basada en agentes inteligentes para sistemas distribuidos de bajo costo.

Abstract

In this work, we researched the dynamic reconfiguration based on intelligent agents for distributed systems of low cost, which consists of avoiding the stoppage of a process, through the coordination of the agents. The agents are software entities, the same ones that were developed in Python, a programming language oriented to open source and high-level objects, whose syntax is easy to interpret, characteristics that offer great potential in the development of agents. Both the main agent and the backup agent are running on two Raspberry Pi 3 model B. This research studies the behavior of dynamic reconfiguration applied to discrete processes, and the case study is the trajectory resolution of a planar Robotic arm of two degrees of freedom. The Robotic arm was implemented, which consists of two servomotors as actuators, and 4 links of 10cm in length. The servomotors receive the information of the angles, a mega Arduino, the same one that has an Ethernet shield W5500 for the communication in a network with the agents. The angles that the servomotors must turn are based on the study of inverse kinematics. To verify the reconfiguration process, different trajectories were evaluated, in which the main agent is subjected to failure, evidencing the behavior of the backup agent, which is satisfactory, since in any failure it restarts the process until its completion with an average of 0.7 seconds. In the statistical analysis of variance (ANOVA), Fisher's calculated value is 0.4835, while Fisher's critical value is 3.2389, so the null hypothesis is accepted. Linux is recommended as an operating system for Raspberry because it is open source.