EXPLOSIVE PERCOLATION BASED ACTIVE SLAM EXPLORATION VIA LIDAR IN UNSTRUCTURED MAP

Date

2023-8-10

Author

Yıldız, Doğan

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

212
views

163
downloads

Cite This

This thesis proposes a novel exploration technique for simultaneous localization and mapping (SLAM) in highly unstructured disaster regions for search and rescue (SAR) operations. As is known to all, disaster regions are very dangerous environments for humans to navigate and search for victims. For this reason, robotics applications in SAR operations gain considerable attention for the last decade. In a disaster region searching for survivors with robots needs finding a safe continuous path to reach the victim's location, and extracting the map of the unstructured area to be shared with SAR teams. However, the highly complex structure of rubbles, the GPS-denied region, and the unpredictable nature of the environment make the process difficult for such applications. Also, the unstable characteristic of debris causes the collapse of certain areas of the disaster region, and this brings unwanted dead-end occurrences making the search operation even harder than before. Unfortunately, even if the distinct progress of SLAM applications in the robotic field, these problems still preserve their effectiveness and they are waiting to be answered. In our proposed method we address these problems and offer an efficient and robust solution with explosive percolation-based exploration technique combined with the active SLAM approach for unstructured environments. The active SLAM approaches mainly focus on the minimization of the total entropy of localization and mapping of the environment. However, in our method, we need to consider not only the entropy values but our main goal of finding the survivors’ location for active exploration. Moreover, during the search finding a continuous path by mapping the disaster environment is affected severely by successive dead-ends. To tackle this problem, we offer a fluid behavior-based approach Explosive Percolation (EP). With the power of this method, the robot can navigate within a complex unstructured environment without being trapped by dead-ends. The main contributions of the thesis to literature can be listed as; finding a novel path-search algorithm based on Explosive Percolation and combining this method with the SLAM approach to obtain a continuous path in a disaster area without being interrupted by dead-ends.

Subject Keywords

Explosive Percolation, SLAM, Localization, Mapping, Search and Rescue

URI

https://hdl.handle.net/11511/105135

Collections

Graduate School of Natural and Applied Sciences, Thesis