In underground, underwater and indoor environments, a robot has to rely solely on its on-board sensors to sense and understand its surroundings. This is the main reason why SLAM gained the popularity it has today. In recent years, we have seen excellent improvement on accuracy of localization using cameras and combinations of different sensors, especially camera-IMU (VIO) fusion. Incorporating more sensors leads to improvement of accuracy,but also robustness of SLAM. However, while testing SLAM in our ground robots, we have seen a decrease in performance quality when using the same algorithms on flying vehicles.We have an additional sensor for ground robots which under the assumptions that the robot moves on a plane surface and slippage of wheels is minimal, achieves high accuracy. These assumptions are usually not entirely accurate, leading to a higher rate of errors when the assumptions do not hold. However, our robot carries a ground penetration radar which will be mostly used to detect metal objects on the floors of buildings, meaning these are good assumptions, because most of the times the floors are plane. In this work we propose a fusion system between Camera, IMU and Encoder as well as a slippage detection algorithm that will avoid fusion of encoder data whenever we have slippage. This way we expect that when the assumptions above hold, the higher accuracy of the encoder will improve the localization,leading to an overall improvement of SLAM. We use modern approaches like loop-closing and optimization to solve the SLAM problem. After the improvement of the pose estimation using SLAM and sensor fusion, we generate a dense map of the environment in addition to sparse maps that ORB SLAM outputs. These maps can be combined with 3D underground maps generated by GPR data. Finally, we generate occupancy maps that are key for autonomous robot navigation.
","abstract_html":"<p>In underground, underwater and indoor environments, a robot has to rely solely on its on-board sensors to sense and understand its surroundings. This is the main reason why SLAM gained the popularity it has today. In recent years, we have seen excellent improvement on accuracy of localization using cameras and combinations of different sensors, especially camera-IMU (VIO) fusion. Incorporating more sensors leads to improvement of accuracy,but also robustness of SLAM. However, while testing SLAM in our ground robots, we have seen a decrease in performance quality when using the same algorithms on flying vehicles.We have an additional sensor for ground robots which under the assumptions that the robot moves on a plane surface and slippage of wheels is minimal, achieves high accuracy. These assumptions are usually not entirely accurate, leading to a higher rate of errors when the assumptions do not hold. However, our robot carries a ground penetration radar which will be mostly used to detect metal objects on the floors of buildings, meaning these are good assumptions, because most of the times the floors are plane. In this work we propose a fusion system between Camera, IMU and Encoder as well as a slippage detection algorithm that will avoid fusion of encoder data whenever we have slippage. This way we expect that when the assumptions above hold, the higher accuracy of the encoder will improve the localization,leading to an overall improvement of SLAM. We use modern approaches like loop-closing and optimization to solve the SLAM problem. After the improvement of the pose estimation using SLAM and sensor fusion, we generate a dense map of the environment in addition to sparse maps that ORB SLAM outputs. These maps can be combined with 3D underground maps generated by GPR data. Finally, we generate occupancy maps that are key for autonomous robot navigation.</p>","abstract_has_math":false,"creators":["Hoxha, Ejup"],"institution":null,"degree_name":"Master of Science (M.S.)","degree_level":"Thesis","degree_discipline":"Computer Science","degree_department":null,"school":null,"contributors":["Jizhong Xiao"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2020,"date_issued":"2020-01-01T08:00:00Z","date_published":"2020-01-01T08:00:00Z","updated_at":"2026-07-24T01:57:28Z","subjects":["Computer","Vision","Epipolar","Geometry","Mapping","Kalman","Fusion","Sensor","Localization","Algebra.","Graphics and Human Computer Interfaces","Probability","Programming Languages and Compilers","Robotics","Software Engineering","Theory and Algorithms"],"languages":[],"rights":[],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://academicworks.cuny.edu/cc_etds_theses/945","outbound_label":"Repository record","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Jizhong Xiao"]},{"key":"dc:creator","label":"Author","values":["Hoxha, Ejup"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.available","label":"Dc Date Available","values":["2020-12-04T08:00:00Z"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Computer Science"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Master of Science (M.S.)"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Computer","Vision","Epipolar","Geometry","Mapping","Kalman","Fusion","Sensor","Localization","Algebra.","Graphics and Human Computer Interfaces","Probability","Programming Languages and Compilers","Robotics","Software Engineering","Theory and Algorithms"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://academicworks.cuny.edu/cc_etds_theses/945"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["In underground, underwater and indoor environments, a robot has to rely solely on its on-board sensors to sense and understand its surroundings. This is the main reason why SLAM gained the popularity it has today. In recent years, we have seen excellent improvement on accuracy of localization using cameras and combinations of different sensors, especially camera-IMU (VIO) fusion. Incorporating more sensors leads to improvement of accuracy,but also robustness of SLAM. However, while testing SLAM in our ground robots, we have seen a decrease in performance quality when using the same algorithms on flying vehicles.We have an additional sensor for ground robots which under the assumptions that the robot moves on a plane surface and slippage of wheels is minimal, achieves high accuracy. These assumptions are usually not entirely accurate, leading to a higher rate of errors when the assumptions do not hold. However, our robot carries a ground penetration radar which will be mostly used to detect metal objects on the floors of buildings, meaning these are good assumptions, because most of the times the floors are plane. In this work we propose a fusion system between Camera, IMU and Encoder as well as a slippage detection algorithm that will avoid fusion of encoder data whenever we have slippage. This way we expect that when the assumptions above hold, the higher accuracy of the encoder will improve the localization,leading to an overall improvement of SLAM. We use modern approaches like loop-closing and optimization to solve the SLAM problem. After the improvement of the pose estimation using SLAM and sensor fusion, we generate a dense map of the environment in addition to sparse maps that ORB SLAM outputs. These maps can be combined with 3D underground maps generated by GPR data. Finally, we generate occupancy maps that are key for autonomous robot navigation.
"]},{"key":"dc:title","label":"Title","values":["V-SLAM and Sensor Fusion for Ground Robots"]}]}],"canonical_facts":{"dc:contributor":["Jizhong Xiao"],"dc:creator":["Hoxha, Ejup"],"dc:date.available":["2020-12-04T08:00:00Z"],"dc:description.abstract":["In underground, underwater and indoor environments, a robot has to rely solely on its on-board sensors to sense and understand its surroundings. This is the main reason why SLAM gained the popularity it has today. In recent years, we have seen excellent improvement on accuracy of localization using cameras and combinations of different sensors, especially camera-IMU (VIO) fusion. Incorporating more sensors leads to improvement of accuracy,but also robustness of SLAM. However, while testing SLAM in our ground robots, we have seen a decrease in performance quality when using the same algorithms on flying vehicles.We have an additional sensor for ground robots which under the assumptions that the robot moves on a plane surface and slippage of wheels is minimal, achieves high accuracy. These assumptions are usually not entirely accurate, leading to a higher rate of errors when the assumptions do not hold. However, our robot carries a ground penetration radar which will be mostly used to detect metal objects on the floors of buildings, meaning these are good assumptions, because most of the times the floors are plane. In this work we propose a fusion system between Camera, IMU and Encoder as well as a slippage detection algorithm that will avoid fusion of encoder data whenever we have slippage. This way we expect that when the assumptions above hold, the higher accuracy of the encoder will improve the localization,leading to an overall improvement of SLAM. We use modern approaches like loop-closing and optimization to solve the SLAM problem. After the improvement of the pose estimation using SLAM and sensor fusion, we generate a dense map of the environment in addition to sparse maps that ORB SLAM outputs. These maps can be combined with 3D underground maps generated by GPR data. Finally, we generate occupancy maps that are key for autonomous robot navigation.
"],"dc:identifier":["https://academicworks.cuny.edu/cc_etds_theses/945"],"dc:subject":["Computer","Vision","Epipolar","Geometry","Mapping","Kalman","Fusion","Sensor","Localization","Algebra.","Graphics and Human Computer Interfaces","Probability","Programming Languages and Compilers","Robotics","Software Engineering","Theory and Algorithms"],"dc:title":["V-SLAM and Sensor Fusion for Ground Robots"],"thesis:degree_discipline":["Computer Science"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["Master of Science (M.S.)"]},"updated_at":"2026-07-24T01:57:28Z"}