Simultaneous Localization And Mapping (SLAM) is the problem of creating a map and localizing an agent at the same time. Several entities have implemented solutions to this problem, but most of them lack flexibility in the use of certain components i.e. constraint generation, feature extraction, graph optimisation, outlier detection, etc.
The aim of this project is to investigate how to define, extract and interface these components of SLAM algorithms in state of the art implementations like RTAB-map, Cartographer and others. Modularity can lead to more adaptable, easier-to-use SLAM solutions for a wider variety of platforms with custom sensors.
Component-based SLAM in RTAB-map
Finished: 2021-10-14
MSc assignment