| Summary: | The field of artificial intelligence has now become part of our daily lives. AI applications range from very simple recommendations to the user to complex autonomous driving systems and now we also have intelligent systems that can have dialogues with human beings such as the GPT-3 system. Moreover, the rapid evolution in computing power and efficiency has allowed the use of neural networks models on mobile devices and generally in low energy consumption devices. In addition to that, if the computational requirements exceed the capabilities of the local processor, it is now possible to perform the calculation on remote computers as long as the delay in transferring the information does not present a problem in the application. This thesis presents a comprehensive target detection and monitoring system using an unmanned aerial vehicle. This system uses the YOLO-V4 object detection model and the DeepSORT identification correlation code. This system has been named SkyTrack and is validated in simulations and in the real world. Initially, the simulation environment where ROS, GAZEBO and an Iris vehicle are used with some modifications is presented. In these tests there is also obstacle avoidance logic by using the ORCA algorithm. Then tests are done in the real world using the DJI Phantom 3 and MINI 2 unmanned aerial vehicle where an android application has been built and the results are positive. Finally, after the results of the experiments have been presented along with points of comparison and optimization, reference is made to open issues of the system and to possible solutions that will be implemented in the future. Ο τομέας της τεχνητής νοημοσύνης έχει πλέον γίνει μέρος της καθημερινότητας μας. Οι εφαρμογές της τεχνητής νοημοσύνης ξεκινούν από πολύ απλές συστάσεις προς τον χρήστη έως και περίπλοκα συστήματα αυτόνομης οδήγησης ενώ πλέον έχουμε και ευφυή συστήματα που μπορούν να έχουν διαλόγους με ανθρώπινα όντα όπως το σύστημα GPT-3. Επίσης η ραγδαία εξέλιξη σε υπολογιστική ισχύ αλλά και σε αποδοτικότητα ...
|
|---|