Robot control using single view depth estimation method

In this thesis formation control in a multi-robot system using single view depth estimation method and circling around a target for mobile robots by feedback linearization were studied. In order to sense objects in three dimension stereo vision is necessary. In other words, two eyes for humans and two cameras for robots are needed for three dimensional sensing. For measuring distance a feature of an object, which is height in this study, is known and this information is used to explain how to measure an object's distance by single view depth estimation method that's developed. Khepera 3 mini robots were used in the experiments and since they were not equipped with the necessary equipments like webcams, servo motors, etc. , these parts were chosen appropriately and integrated to the robots. A driver for chosen webcam was also compiled. OpenCV was used for image processing. Necessary OpenCV libraries were cross-compiled and added on robots via a flash disk. In these studies, which are done on 3 or 4 robots, it was observed that robots can successfully go to their destination points without breaking formation, using their designed controllers and a simple webcam mounted on them. In the study using a single robot using the feedback linearization techniques a controller for circling around a target with a predefined radius was implemented. This study constitutes a preliminary study towards a multi-agent system application in which robots constitutes targets to each other.