Vision object tracking by color with quadcopter

Abstract

The project aims to investigate the feasibility of using an unmanned aerial vehicle to track a moving object by its colour. This paper presents the methodology used to design the hardware and software infrastructures to render an integrated system that can fulfil the objectives of this project. The paper can be segmented into a few sections, namely the hardware, the software, and the testing sections. The components discuss under the hardware section will demonstrate how “eyes” are given to the drone, allowing it to “see” the things around it. On the other hand, in the software section, the algorithm and the communication methodology used to equip the drone with the intelligence to process the image it “sees” and return it as where to hover to, will be discussed. After the hardware and software are developed, the integration process will be implemented, and the final tests will be conducted. Due to the dangerous nature of conducting tests with a drone, it is always recommended to conduct experiments on a simulator, before testing it out on an actual drone. By doing this, it reduces the chances of any possible damages caused in case the algorithm fails and the drone crashes. Adding on, running tests on a simulation will provide the developer insights on the potential problems that may exists in the algorithm, making it easier for the developer to troubleshoot.