Multi-objective approach for robot exploration

Abstract

This work addresses the problem of single robot coverage and exploration of an environment with the aim of nding a specic previously known object. As limited time is a constraint of interest we cannot search for an innite number of points. Thus, we proposeto nd good points (also called search points) to place the robot sensors in order to acquire information from the environment and nd the desired object. Given the interesting properties of the Generalized Voronoi Diagram (GVD), we dene the search points along this roadmap. We redene the problem of nding these search points as a multi-objectiveoptimization one. NSGA-II is used as optimizer and ELECTRE I is applied as a decision making tool. We also solve a Chinese Postman Problem to optimize the path followed by the robot in order to visit the computed search points. To identify the desired object in environment, we used a fast and robust object recognition application which is called Speeded Up Robust Features (SURF) algorithm. Simulations on Stage with implementation in ROS are also presented. The proposed approach tested on an real robot in a real world situation that indicates the applicability of our method. Lastly, statistical analysis shows a comparison between the solution found by our methodand two others.