In English

Investigating Simultaneous Localization and Mapping for AGV systems

Albin Pålsson ; Markus Smedberg
Göteborg : Chalmers tekniska högskola, 2017. 72 s.
[Examensarbete på avancerad nivå]

The purpose of this project is to investigate solutions for Simultaneous Localization and Mapping (SLAM) in the context of Automated Guided Vehicles (AGVs). This thesis presents implementation details of a prototype system for AGVs, which was developed with the intention of allowing application-specific testing of SLAM. Three of the most prominent open-source SLAM algorithms, available in ROS, have been evaluated and critically compared. Furthermore, basic background and explanation of the critical problems of SLAM are presented. The SLAM algorithms have been evaluated based on resulting map quality as well as resource requirements. Map quality is tested based on both visual comparison with a ground truth and the correctness of estimated distances in the map. In addition to this, results are presented from tests which have been conducted in order to test the SLAMgenerated maps in AGV application areas. This includes tests where critical tasks, such as the robot’s precision while navigating with the use of a SLAM-generated map, has been assessed. The presented prototype system is based on the Robot Operating System (ROS), which includes state-of-the-art libraries and tools for robotic navigation. The prototype enables testing of navigation and mapping software available in ROS by publishing sensor data, from a physical AGV. The implemented software publishes readings from a laser range scanner, wheel encoders and a gyroscope. These sensor reading are published in a standardized way, making them accessible to applications in the ROS framework. The presented results from the tests answers the question of whether or not SLAM is suitable in the intended environment.

Nyckelord: AGV, Navigation, Robot Operating System, ROS, Automation, SimultaneousLocalization and Mapping, SLAM, NDT

Publikationen registrerades 2017-06-22. Den ändrades senast 2017-06-22

CPL ID: 250073

Detta är en tjänst från Chalmers bibliotek