The ability to perceive the environment and to localize in it are essential for the robots to perform their assigned tasks. This ability is called simultaneous localization and mapping (SLAM). Many algorithms have been deployed to solve the SLAM problem. Some of them have been implemented on Robot Operating System (ROS), a framework for writing robotic software. This thesis provides a review of some 2D SLAM algorithms which are implemented in ROS. The algorithms to be reviewed are GMapping, Google Cartographer, SLAM Toolbox, and Hector SLAM. To create the review, several experiments are conducted. These experiments test the algorithms on several datasets with various parameter configurations for each algorithms. The resulting maps and the performance of each algorithm during the runtime are discussed. Our main observations are related to the effect of each parameter on each algorithm and the difference across the algorithms especially when the algorithms seem to use same components, e.g. motion filter, but give different results. These can provide insights in tuning the parameters for each algorithm and can help in deciding which algorithms to use in the intended application.
La capacità di percepire l'ambiente e di localizzarsi in esso è essenziale per i robot per eseguire i compiti assegnati. Questa capacità è chiamata simultaneous localization and mapping (SLAM). Molti algoritmi sono stati implementati per risolvere il problema SLAM. Alcuni di essi sono stati implementati su Robot Operating System (ROS), un framework per la scrittura di software robotico. Questa tesi fornisce una revisione di alcuni algoritmi SLAM 2D che sono implementati in ROS. Gli algoritmi da rivedere sono GMapping, Google Cartographer, SLAM Toolbox e Hector SLAM. Per creare la recensione, vengono condotti diversi esperimenti. Questi esperimenti testano gli algoritmi su diversi set di dati con varie configurazioni di parametri per ciascun algoritmo. Vengono discusse le mappe risultanti e le prestazioni di ciascun algoritmo durante il runtime. Le nostre osservazioni principali sono correlate all'effetto di ciascun parametro su ciascun algoritmo e alla differenza tra gli algoritmi, specialmente quando gli algoritmi sembrano usare gli stessi componenti, ad es. filtro di movimento, ma danno risultati diversi. Questi possono fornire approfondimenti sull'ottimizzazione dei parametri per ciascun algoritmo e possono aiutare a decidere quali algoritmi utilizzare nell'applicazione prevista.
A review of 2D SLAM algorithms on ROS
LUKNANTO, BAYU KANUGRAHAN
2018/2019
Abstract
The ability to perceive the environment and to localize in it are essential for the robots to perform their assigned tasks. This ability is called simultaneous localization and mapping (SLAM). Many algorithms have been deployed to solve the SLAM problem. Some of them have been implemented on Robot Operating System (ROS), a framework for writing robotic software. This thesis provides a review of some 2D SLAM algorithms which are implemented in ROS. The algorithms to be reviewed are GMapping, Google Cartographer, SLAM Toolbox, and Hector SLAM. To create the review, several experiments are conducted. These experiments test the algorithms on several datasets with various parameter configurations for each algorithms. The resulting maps and the performance of each algorithm during the runtime are discussed. Our main observations are related to the effect of each parameter on each algorithm and the difference across the algorithms especially when the algorithms seem to use same components, e.g. motion filter, but give different results. These can provide insights in tuning the parameters for each algorithm and can help in deciding which algorithms to use in the intended application.File | Dimensione | Formato | |
---|---|---|---|
A Review of 2D SLAM Algorithms on ROS.pdf
accessibile in internet per tutti
Descrizione: Thesis report revision 1
Dimensione
31.38 MB
Formato
Adobe PDF
|
31.38 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/164687