Supervisors info:
Ιωάννης Καλατζής, Αναπληρωτής Καθηγητής, Μηχανικών Βιοϊατρικής, Πανεπιστήμιο Δυτικής Αττικής
Summary:
Nowadays, more and more facilities, such as warehouses, hospitals and hotels, manage the transportation of their objects within their premises using robotic vehicles. These vehicles are equipped with special cameras and sensors and can either map and navigate on site or follow markings, mainly on the floor, and navigate through them. The present thesis attempts to create and construct an integrated tracking and autonomous vehicle driving system using image analysis techniques solely.
Firstly, the general structure of the system is presented, which consists of three stages: the autonomous vehicles, the central server and the administrator. Specific software was developed for the communication between the three stages, where the server plays the most substantial role. In general, once the users of the system have been identified, the communication between them is based on sending messages to the server and then forwarding them to the respective recipient. The communication is based on secure, asynchronous and two-way protocols, so that each stage of the system is at any time aware of the status of the other stages.
In addition, with respect to the operation of the system, the administrator may assign routes to the vehicles by means of the server, whereafter the vehicles can navigate from the starting node to the destination node moving through a dedicated space with markings on the floor and using exclusively image analysis techniques. The special area includes crossroads, which the vehicles have to identify and, using their internal map, they decide the required travel direction in order to reach their destination.
Finally, a test of the navigation system’s operation was carried out in a laboratory environment under non-optimal conditions. Initially, the administrator selected and assigned a desired route to one vehicle, so that the latter started its navigation. During this test, it was examined whether the communication between the operator, the server and the vehicle as well as the vehicle's vision and navigation algorithms were correct and successful, since the vehicle had to pass through different types of crossroads and under poor lighting conditions. It was found that the communication was carried out smoothly and as expected and that the vehicle reached the desired destination. However, the simplicity of the system as well as its limitations make further study and development necessary. In conclusion, the developed system of the present study is considered an important basis for the development of more complex tracking and autonomous vehicle driving systems using image analysis techniques solely.
Keywords:
ROS, robotic vehicles, autonomous vehicles, autonomous driving, OpenCV, image processing, tracking systems, embedded systems, Arduino, microcontroller, ESP8266
