Vision based Systems for Localization in Service Robots

Abstract

This chapter proposed three vision based methods to localize objects and to determine the orientation of robots. Active and passive stereo visions along with neural networks prove to be efficient techniques in localization of objects and robots. The unified stereo vision system discussed in this chapter depicts the human biological stereo vision system to recognise and localize objects. However a good database of typical situations is necessary is implement the system. Combining vision sensors with ultrasonic sensors is an effective method to combine the information from both sensors to localize objects. All the three vision systems were tested in real-time situations and their performance were found to be satisfactory. However the methods proposed are only suitable for controlled indoor environments, further investigation is required to extend the techniques to outdoor and challenging environments