Evaluation of a Fault-tolerant Model for Tactic Operations of Mobile Robotic Groups Using Genetic Algorithms

Gustavo Pessin,Fernando Osório

doi:10.19153/cleiej.13.3.3

Abstract

   This paper addresses the evaluation of a fault-tolerant model for tactic operations of mobile robotic groups. The coordinated action of the group is planned with Genetic Algorithms (GA) and aims to act on an environmental disaster scenario, simulated as a forest fire. The robotic squad should surround the fire and avoid fire's propagation. Initially, we evaluate several parameters in the GA, seeking to obtain the set of parameters that would accomplish the more efficient evolution. Then, we simulate failures in robots operations in order to evaluate strategies of reorganization. The simulation's results1 showed that with an adequate set of parameters it is possible to get satisfactory strategic positions to coordinate and to reorganize the robotic group in case of robot failures.  

Highlights

The continuous evolution provided by mobile robotics research area has made even more efficient robots for several functions
Research about controlling complex motor functions are developed on several research centers around the world, encompassing studies about sensors and actuators, positioning, navigation and localization in addition to many other requirements related to robotic hardware, as demonstrated by [1] and [2]
In [8] we proposed a Genetic Algorithm to accomplish the formation of a robotic squad that should perform a firefighting task; the previously proposed Genetic Algorithms (GA) did not deal with reorganization in case of robot failures

Summary

INTRODUCTION

The continuous evolution provided by mobile robotics research area has made even more efficient robots for several functions. These work described shows that the application of mobile robotics in control of incidents is an important and active topic of research and development These several competitions show that there is not yet a definitive or more adequate solution to the problem, and it is an open research field. These works present acceptable results for static environments; on the other side [22] describes a possible solution for operation in dynamic places, where the robot perform the navigation using GA This robot is equipped with obstacle sensors and when identifies a possible collision, it stops and executes again the planning module using GA. The GA tries to minimize the fitness function value, which means less burned vegetation, less created firebreaks, and less difference between the sizes of firebreaks of each robot

EXPERIMENTS AND RESULTS

FUTURE WORK

CONCLUSION