diplomsko delo
Mihael Šinkec (Author), Danijel Skočaj (Mentor)

Abstract

Cilj diplomske naloge je razviti rešitev za usmerjanje kvadrokopterja s pomočjo telesnih gest. Rešitev smo implementirali v obliki mobilne aplikacije za operacijski sistem Android. Uporabili smo kvadrokopter podjetja Ryze Robotics, ki ponuja omrežni vmesnik, preko katerega ga je možno usmerjati ter pridobivati video tok iz njegove vgrajene kamere. Naša implementacija je strukturirana kot zaprta zanka, ki je zgrajena iz treh zaporednih faz oz.\ modulov --- dekodiranja in pretvorbe slike v ustrezen format, ocene poze telesa ter klasifikacije geste. Na podlagi rezultata tega cevovoda se določi akcija za kvadrokopter, ki jo mora ta izvesti. Delovanje končne rešitve smo evalvirali z merjenjem latenc posameznega dela zanke, kot tudi empirično na dejanskih primerih ukazovanja z gestami. Pokazali smo, da je naš sistem zmožen upravljanja kvadrokopterja v praksi. Izpostavili smo posamezne težave, za katere smo predlagali možne rešitve. Prvi tip težav se pojavi zaradi raznih nejasnosti na vhodnih slikah. Te so rešljive s predprocesiranjem slike. Ostale težave so povezane z arhitekturo modela za klasifikacijo gest, ki ni optimalna za naš problem.

Keywords

kvadrokopter;geste;računaniški vid;mobilna aplikacija;robotika;strojno učenje;računalništvo;računalništvo in informatika;visokošolski strokovni študij;diplomske naloge;

Data

Language: Slovenian
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UL FRI - Faculty of Computer and Information Science
Publisher: [M. Šinkec]
UDC: 004:629.014.9(043.2)
COBISS: 31566339 Link will open in a new window
Views: 1449
Downloads: 156
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Drone control using gestures
Secondary abstract: The goal of this thesis is to develop a solution for controlling a drone by using body gestures. We developed the solution in the form of a mobile application for Android devices. We used a Tello drone, produced by the company Ryze Robotics. This drone offers a network interface for receiving commands. It also has a built-in camera for capturing a video feed, which can be streamed to a connected device. Our implementation is structured as a closed loop, which contains three consecutive phases or modules. These are image decoding, pose estimation and gesture classification. Based on the results of the gesture classifier, a command is defined and issued to the drone. We evaluated the analyzed system by measuring latencies across modules of the loop and also by using empirical methods on actual examples of controlling the drone with gestures. We demonstrated, that our is able to function properly in practice. At the end we expose current issues of our solution and suggest possible improvements. The first group of issues occur because of ambiguities in the input images. These can be resolved by preprocessing the image. The other issues are related to the gesture classifiers architecture, which is not optimal for our problem.
Secondary keywords: drone;gestures;computer vision;mobile application;robotics;machine learning;computer science;computer and information science;diploma thesis;
Type (COBISS): Bachelor thesis/paper
Study programme: 1000470
Embargo end date (OpenAIRE): 1970-01-01
Thesis comment: Univ. v Ljubljani, Fak. za računalništvo in informatiko
Pages: 58 str.
ID: 12037021