Mobile robot trajectory estimation based on graph optimization

Rad se bavi problemom istovremene lokalizacije i izgradnje karte, koji zahtijeva rješenje u većini primjena mobilne robotike. U prvom se poglavlju govori općenito o mobilnoj robotici, o senzorima i gibanju mobilnih robota i o ciklusu mjerenjai gibanja. Drugo poglavlje opisuje SLAM problem, njegovu definiciju pomoću grafa položaja i osnovne metode i algoritme koji se koriste za rješavanje tako kodiranog SLAM problema. Treće poglavlje opisuje dva paketa, Ceres i g2o, koji nude biblioteke funkcija za rješavanje 2Di 3DSLAM problema. U četvrtom se poglavlju Ceres i g2o testiraju na šest 2D ulaznih datoteka i šest 3D ulaznih datoteka i rezultati testiranja se uspoređuju.The thesis deals with the problem of simultaneous localization and map construction, which requires a solution in most applications of mobile robotics. The first chapter discusses mobile robotics in general, sensors and motion of mobile robots, and the cycle of observation and motion. The second chapter describes the SLAM problem, its coding using a pose graph, and the basic methods and algorithms used to solve such SLAM problems. The third chapter describes two packages, Ceres and g2o, which offer function libraries for solving 2D and 3D SLAM problems. In the fourth chapter, Ceres and g2o are tested on six 2D input files and six 3D input files and the test results are compared