Investigation of slam and vslam algorithms, development and application of a new map merging technique

The solution of the Simultaneous Localization and Mapping (SLAM) problem is the basis of many advanced pieces of research used in today's technology. Considering the importance of the topic, both classical and modern period SLAM algorithms have been investigated in this dissertation. Various applications which compare the existing techniques and different sensors have realized for classical period SLAM algorithms. As well as the investigation of the existing modern period SLAM methods, an improvement which basically transforms the ORB-SLAM technique feature maps into the density-based maps is provided and the results are demonstrated with a real-time application. In addition to these, studies on active SLAM and map merging which are addressed in the neighborhood of SLAM have been realized. Two new approaches have been proposed for the navigation part of active SLAM algorithms. A deep learning-based method is developed for the first study and an algorithm framework called left-oriented is enhanced for the second study. The developed algorithms and their performance are examined on simulation and real-time environments. The failure of the current methods on environments which have symmetrical features has been witnessed for the map merging problem of the multi-robot. Based on this failure, a trusted feature and related semantic map merging algorithm have been proposed within the scope of this dissertation. This developed method is analyzed by simulation and real-time experiments and advantages of the method over the existing methods have been observed. Besides all these, localization/mapping applications on humans other than robots have been examined. Real-time tests are performed with an inertial measurement unit (IMU) mounted on the pedestrian foot. The problems with unknown initial points are handled. Discussions and development on how to find the initial position and coordinate matching between IMU and the global map have been carried out by various real-time experiments.