Implementation of Video-based Person Tracking in a Drone System

Detta är en M1-uppsats från KTH/Skolan för kemi, bioteknologi och hälsa (CBH)

Sammanfattning: Technological instruments are used in sports to record and analyze data of performing athletes, in order to improve techniques and increase competitiveness. A method to allow for recording of data of a moving subject is by usage of a depth camera mounted on a drone that can track and follow a subject. The objective was to develop a drone system that was capable of autonomous operation based on follow-me mode. The original prototype lacked required hardware for reliable orientation in environments with varying air-pressure, and software for a follow-me mode-system. Height above ground was measured using a downward facing depth sensor and a 3D image of the subject was generated by a depth camera. The data was then used by the drone to navigate through the environment. Overall performance of the height adjustment would be sufficient to allow for autonomous operation according to test results. Testing of the follow-me mode showed that the current configuration was not capable of retaining a sufficiently consistent position relative to the subject, hence high video quality was not achieved. However, it gave a positive indication that autonomous operation based on a depth camera is possible. The concept has a high potential and if the system would be further developed it could allow athletes to record and analyze data in order to improve techniques and increase competitiveness.

  HÄR KAN DU HÄMTA UPPSATSEN I FULLTEXT. (följ länken till nästa sida)