MULTI-QUADROTOR DESIGN FOR MAPPING IN SEARCH AND RESCUE OPERATION

• Main Author: Renardi, Bernard
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/42024
• Institution: Institut Teknologi Bandung

Natural disasters often occur in Indonesia, causing loss of lives and widespread destruction such as changes of contour. This change will impede the SAR (Search and Rescue) process because of the difficul field and unusable map of the area. To solve this problem, quadrotor will be used to create a new map of the area and identified victims by using image processing. But, the use of only one quadrotor won’t be enough to cover the whole disaster area because of the battery limit. This research will introduce a multi-quadrotor system using Robot Operating System (ROS) as the framework for the software development. The designed system is an integration of two ROS package, which is ardrone_autonomy and tum_ardrone. The multi-quadrotor will take images of contour using a 720p front camera with frequency of 2 – 4 Hz while flying. The images taken with a needed specification will be processed into a new map of the area using Agisoft Photoscan. This new map will be processed by a YOLO-based object detection algorithm for victim identification purpose. In this research, a multi-quadrotor system has been designed and implemented using AR. Drone 2.0 in some flights. Image processing using Agisoft Photoscan and YOLO algorithm is also succesfully done and results in high resolution images of map.