AN EFFICIENT OBSTACLE RECOGNITION SYSTEM FOR HELPING MOBILE ROBOT NAVIGATION

摘要

This paper proposes a novel efficient obstacle recognition system for helping the navigation of mobile robots, which is designed to operate in connection with agent-based control architectures (its insertion in the whole agent-based control system is exemplified). The capability of recognizing objects allows difining more complex behaviors to be programmed, as well as the associatin of different behaviors to distinct objects recognized by the robot. A single CCD camera and a group of ultrasonic transducers responsible for activating the camera whenever the robot gets close to any obstacle compose the system hete proposed, whose operation is characterized as a 2 1/2-D vision system. Although being simple and cheap,this system allows recognizing a set of meaningful objects present in the robot workspace,even when more than one object is present in the scene, like it is shown in the examples presented. A main feature regarding this system is also that all the sensorial data are processed onboard the robot, and demands non-intensive computation for interpreting the sensorial information. Finally, a control system including the recognition system here proposed is shown,whose objective is to guide the robot outside the lab, through a door the system recognizes.