Application of Vision Target Localization for Mobile Robot

Abstract

A though challenge for an autonomous mobile robot is reliable targets detection and localization. Many types of sensor are used for that purposes such as infrared, sonar, vision and laser. Monocular vision is one of the methods used for that reasons. Based on current trends vision sensor is used as different functions such as target recognition, obstacles avoidance, and navigation. This paper addresses issue on targets localization using a single color vision system. The objective is to localize targets at various locations to be retrieved by the mobile robot. The distances towards the targets are estimated based on decreasing in their sizes as the locations are farther from the mobile robot. Targets detection is done using color thresholding technique in RGB color space. The threshold value is selected by taking mean value of image samples and its standard deviation of the color components difference. The simple features such as diameter, x-y ratio and area are used to verify the correct targets are detected. The localization method takes less than one second to be completed where the system is running on Windows platform. The efficiency of the estimation is more than 90 percents and targets used in the testing are golf balls.