Upravljanje smjerom gibanja mobilnoga robota zasnovano na vizualnim značajkama

master's thesis

• Title (english): Mobile robot heading control based on visual features
• Author: Viktor Prusac
• Mentor: Ivan Marković (mentor)
• Committee member: Ivan Marković (predsjednik povjerenstva)
• Committee member: Ivan Petrović (član povjerenstva)
• Committee member: Branimir Novoselnik (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-02-19, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Ovaj rad rješava problem lateralnog odstupanja robota koje dolazi kod pravocrtnog
gibanja robota sa zadanim smjerom. Do odstupanja dolazi zbog proklizavanja rob-
ota, nepreciznosti kotača, motora ili baterije ili vanjskih fizičkih čimbenika. Ovaj
rad rješava taj problem koristeći vizualne značajke dobivene od kamere. Rješenje je
napisano u programskom jeziku Python, na mobilnom Husky robotu u ROS okruženju
koje nudi potpunu integraciju s Husky robotom i mnogim korisnim alatima kao što su
Gazebo simulator ili RViz vizualizator. Iako je eksperimentalni dio rada demonstriran
u simulatoru s različitim hiperparametrima, na različitim mapama i na Husky robotu,
rješenje je primjenjivo na fizičku okolinu i na sve mobilne robote koji podržavaju ROS.
Rješenje implementira obradu slike u stvarnom vremenu pomoću Lucas-Kanade al-
goritma optičkog toka. Taj algoritam pronalazi značajke sa slike te traži njihov po-
mak u odnosu na sljedeću dobivenu sliku. Iz tih značajki se izračunava točka Fokusa
ekspanzije (FOE), koja predstavlja nepomičnu točku na epipolarnoj liniji vektora op-
tičkih tokova značajki. Kako robot lateralno odstupa tako se i FOE mijenja. Uzimajući
to u obzir, pronalazi se smjer te veličina lateralnog odstupanja robota te se izračuna
nova kutna brzina koja se šalje robotu kako bi ispravio putanju i eliminirao lateralno
odstupanje.

Abstract (english)

This thesis solves the problem of lateral drifting of the robot that occurs when the
robot moves in a straight line with a given direction. Drifting occurs due to robot
slippage, wheel, motor or battery inaccuracies, or external physical factors. This pa-
per solves this problem using visual features obtained from the camera. The solution
is written in Python programming language, on a mobile robot Husky in the ROS
framework that offers full integration with Husky robot and many useful tools such as
Gazebo simulator or RViz visualizer. Although the experimental part of he work was
demonstrated in a simulator with different hyperparameters, on various maps and on
a Husky robot, the solution is also applicable to the physical environment and to all
mobile robots that support ROS. The solution implements real-time image processing
using the Lucas-Kanade optical flow algorithm. This algorithm finds the features from
the image and looks for their offset relative to the next image obtained. Using these
features, the Focus of Expansion (FOE) is calculated, which is a fixed point on the
epipolar line of the feature’s optical flow vector. As the drifting occurs, FOE is also
changing. Taking that into account, the direction and magnitude of the lateral drift of
the robot are found and the new angular velocity is calculated. The velocity is sent the
robot to correct the trajectory and eliminate the lateral drift.

• Keywords: ROS; Husky robot; obrada slike; Lucas-Kanade optički tok; Fokus Ekspanzije
• Keywords (english): ROS; Husky robot; image processing; Lucas-Kanade optical flow; Focus of Expansion
• Language: croatian
• URN:NBN: urn:nbn:hr:168:595429
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 13:35:34