Self-localization of mobile robot based on monocular and extended kalman filter

Self-localization is a fundamental requirement for a mobile robot. Now most of the self-localization methods are based on the environment map, but to build the environment map is no doubt will increase the complexity of the algorithm. Aiming at the indoor environments without prior map, in order to achieve fast and accurate self-localization of the robot, a visual self-localization method based on feature tracking has been studied in this paper. The basic principle of the algorithm is to achieve real-time and robust localization via tracking path features. First the output model of the monocular camera has been changed to HSV (hue, saturation and value) space, and then different hue and saturation thresholds have been set to identify features from the environments. Extended Kalman filter algorithm has been used for merging the features´ parameters into the parameters of the robot´s pose which get from the odometer, and estimating the robot´s new position and orientation, just the location of the robot could be updated. Finally, it is demonstrated by simulation that the algorithm is effective.