{"id":{"repo_id":"uiuc","oai_identifier":"oai:www.ideals.illinois.edu:2142/105713"},"canonical_url":"https://search.dev.ndltd.org/etd/uiuc/oai:www.ideals.illinois.edu:2142/105713","repository":{"repo_id":"uiuc","name":"University of Illinois - Urbana-Champaign","base_url":"https://www.ideals.illinois.edu/oai-pmh"},"display":{"title":"Measurement consensus metric aided lidar localization and mapping","abstract":"In LiDAR Simultaneous Localization and Mapping (SLAM), probability distribution approximations of point clouds, such as the Normal Distribution Transform (NDT), can be leveraged for compact map storage and fast lookup. Augmentation of these techniques with a measurement consensus-based estimate of feature reliability can make them robust to measurement faults. In the presence of a large number of outliers, a nominal error model is required to estimate the measurement consensus. As laser measurement faults are dependent on a large number of operating conditions and a dynamic feature arrangement, the point cloud nominal error model is intractable. In addition to this, errors of clustered low-reliability features can be highly correlated and induce a disproportionate drift in the navigation solution. A measurement consensus metric must be sensitive to the difference in the arrangement of features while quantifying the reliability of a navigation solution. In this work, we propose a consensus-NDT SLAM framework that utilizes measurement consensus metrics to achieve faster convergence and reduce the impact of outliers, feature mismatches, and occlusions on the navigation solution. The measurement consensus metric provides an estimate of the reliability of the navigation solution by a two-tiered quantification approach. First, an NDT approximation is used for the nominal error model on a voxel level. The voxel consensus metric is quantified using a smooth normalized residual squared sum that detects measurement faults. Second, this metric is combined to give the localization consensus metric by calculating the normalized inverse of the Consensus Weighted Dilution of Precision (CWDOP) for all voxels. In this way, the localization consensus metric models the impact of the voxel arrangement on the reliability of the navigation solution. In the LiDAR odometry component of consensus-NDT SLAM, first an intermediate navigation solution is obtained using all features present in the point cloud. After this coarse optimization, low consensus metric voxels are removed and the remaining voxels are used to obtain a fine-tuned navigation solution. The removal of low consensus voxels reduces the impact of outliers, feature mismatches, and occlusions on the obtained navigation solution. Voxel consensus metric thresholding leads to faster convergence and reduced computational load without sacrificing localization accuracy. The mapping component uses the LiDAR odometry solutions as initializations for the optimization and updates the map and LiDAR poses using high consensus voxels. To validate our proposed architecture, a localization and mapping experiment is conducted using real-world data collected on the campus of the University of Illinois at Urbana-Champaign. Results show that consensus-NDT SLAM maintains localization accuracy while providing faster convergence with a lower computational cost. Additional experiments conducted using point clouds from the KITTI Vision Benchmark Suite show that the proposed consensus metric successfully quantifies the goodness of match between the measurement and reference with respect to the navigation solution, irrespective of the algorithm used.","abstract_html":"In LiDAR Simultaneous Localization and Mapping (SLAM), probability distribution approximations of point clouds, such as the Normal Distribution Transform (NDT), can be leveraged for compact map storage and fast lookup. Augmentation of these techniques with a measurement consensus-based estimate of feature reliability can make them robust to measurement faults. In the presence of a large number of outliers, a nominal error model is required to estimate the measurement consensus. As laser measurement faults are dependent on a large number of operating conditions and a dynamic feature arrangement, the point cloud nominal error model is intractable. In addition to this, errors of clustered low-reliability features can be highly correlated and induce a disproportionate drift in the navigation solution. A measurement consensus metric must be sensitive to the difference in the arrangement of features while quantifying the reliability of a navigation solution. In this work, we propose a consensus-NDT SLAM framework that utilizes measurement consensus metrics to achieve faster convergence and reduce the impact of outliers, feature mismatches, and occlusions on the navigation solution. The measurement consensus metric provides an estimate of the reliability of the navigation solution by a two-tiered quantification approach. First, an NDT approximation is used for the nominal error model on a voxel level. The voxel consensus metric is quantified using a smooth normalized residual squared sum that detects measurement faults. Second, this metric is combined to give the localization consensus metric by calculating the normalized inverse of the Consensus Weighted Dilution of Precision (CWDOP) for all voxels. In this way, the localization consensus metric models the impact of the voxel arrangement on the reliability of the navigation solution. In the LiDAR odometry component of consensus-NDT SLAM, first an intermediate navigation solution is obtained using all features present in the point cloud. After this coarse optimization, low consensus metric voxels are removed and the remaining voxels are used to obtain a fine-tuned navigation solution. The removal of low consensus voxels reduces the impact of outliers, feature mismatches, and occlusions on the obtained navigation solution. Voxel consensus metric thresholding leads to faster convergence and reduced computational load without sacrificing localization accuracy. The mapping component uses the LiDAR odometry solutions as initializations for the optimization and updates the map and LiDAR poses using high consensus voxels. To validate our proposed architecture, a localization and mapping experiment is conducted using real-world data collected on the campus of the University of Illinois at Urbana-Champaign. Results show that consensus-NDT SLAM maintains localization accuracy while providing faster convergence with a lower computational cost. Additional experiments conducted using point clouds from the KITTI Vision Benchmark Suite show that the proposed consensus metric successfully quantifies the goodness of match between the measurement and reference with respect to the navigation solution, irrespective of the algorithm used.","abstract_has_math":false,"creators":["Kanhere, Ashwin Vivek"],"institution":"University of Illinois at Urbana-Champaign","degree_name":"M.S.","degree_level":"Thesis","degree_discipline":"Aerospace Engineering","degree_department":null,"school":null,"contributors":["Gao, Grace X"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2019,"date_issued":"2019-11-26T20:35:15Z","date_published":"2019-11-26T20:35:15Z","updated_at":"2026-07-22T22:24:44Z","subjects":["measurement consensus","LiDAR","SLAM"],"languages":["en"],"rights":["Copyright 2019 Ashwin Vivek Kanhere"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/2142/105713","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Gao, Grace X"]},{"key":"dc:creator","label":"Author","values":["Kanhere, Ashwin Vivek"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2019-11-26T20:35:15Z","2019-07-19","2019-08"]},{"key":"dc:type","label":"Dc Type","values":["text"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Aerospace Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["M.S."]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Illinois at Urbana-Champaign"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["measurement consensus","LiDAR","SLAM"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright 2019 Ashwin Vivek Kanhere"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["http://hdl.handle.net/2142/105713"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["In LiDAR Simultaneous Localization and Mapping (SLAM), probability distribution approximations of point clouds, such as the Normal Distribution Transform (NDT), can be leveraged for compact map storage and fast lookup. Augmentation of these techniques with a measurement consensus-based estimate of feature reliability can make them robust to measurement faults. In the presence of a large number of outliers, a nominal error model is required to estimate the measurement consensus. As laser measurement faults are dependent on a large number of operating conditions and a dynamic feature arrangement, the point cloud nominal error model is intractable. In addition to this, errors of clustered low-reliability features can be highly correlated and induce a disproportionate drift in the navigation solution. A measurement consensus metric must be sensitive to the difference in the arrangement of features while quantifying the reliability of a navigation solution. In this work, we propose a consensus-NDT SLAM framework that utilizes measurement consensus metrics to achieve faster convergence and reduce the impact of outliers, feature mismatches, and occlusions on the navigation solution. The measurement consensus metric provides an estimate of the reliability of the navigation solution by a two-tiered quantification approach. First, an NDT approximation is used for the nominal error model on a voxel level. The voxel consensus metric is quantified using a smooth normalized residual squared sum that detects measurement faults. Second, this metric is combined to give the localization consensus metric by calculating the normalized inverse of the Consensus Weighted Dilution of Precision (CWDOP) for all voxels. In this way, the localization consensus metric models the impact of the voxel arrangement on the reliability of the navigation solution. In the LiDAR odometry component of consensus-NDT SLAM, first an intermediate navigation solution is obtained using all features present in the point cloud. After this coarse optimization, low consensus metric voxels are removed and the remaining voxels are used to obtain a fine-tuned navigation solution. The removal of low consensus voxels reduces the impact of outliers, feature mismatches, and occlusions on the obtained navigation solution. Voxel consensus metric thresholding leads to faster convergence and reduced computational load without sacrificing localization accuracy. The mapping component uses the LiDAR odometry solutions as initializations for the optimization and updates the map and LiDAR poses using high consensus voxels. To validate our proposed architecture, a localization and mapping experiment is conducted using real-world data collected on the campus of the University of Illinois at Urbana-Champaign. Results show that consensus-NDT SLAM maintains localization accuracy while providing faster convergence with a lower computational cost. Additional experiments conducted using point clouds from the KITTI Vision Benchmark Suite show that the proposed consensus metric successfully quantifies the goodness of match between the measurement and reference with respect to the navigation solution, irrespective of the algorithm used.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2019-11-26 without embargo terms","The student, Ashwin Kanhere, accepted the attached license on 2019-07-19 at 11:46.","The student, Ashwin Kanhere, submitted this Thesis for approval on 2019-07-19 at 11:52.","This Thesis was approved for publication on 2019-07-19 at 13:35.","DSpace SAF Submission Ingestion Package generated from Vireo submission #14363 on 2019-11-26 at 12:54:11","Made available in DSpace on 2019-11-26T20:35:15Z (GMT). No. of bitstreams: 2 KANHERE-THESIS-2019.pdf: 2494209 bytes, checksum: ba4a6e406fd3cadce50afc5bfb6695de (MD5) LICENSE.txt: 4211 bytes, checksum: 00f29dc20abfd65b5e322bffbff276ed (MD5) Previous issue date: 2019-07-19"]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Measurement consensus metric aided lidar localization and mapping"]}]}],"canonical_facts":{"dc:contributor":["Gao, Grace X"],"dc:creator":["Kanhere, Ashwin Vivek"],"dc:date":["2019-11-26T20:35:15Z","2019-07-19","2019-08"],"dc:description":["In LiDAR Simultaneous Localization and Mapping (SLAM), probability distribution approximations of point clouds, such as the Normal Distribution Transform (NDT), can be leveraged for compact map storage and fast lookup. Augmentation of these techniques with a measurement consensus-based estimate of feature reliability can make them robust to measurement faults. In the presence of a large number of outliers, a nominal error model is required to estimate the measurement consensus. As laser measurement faults are dependent on a large number of operating conditions and a dynamic feature arrangement, the point cloud nominal error model is intractable. In addition to this, errors of clustered low-reliability features can be highly correlated and induce a disproportionate drift in the navigation solution. A measurement consensus metric must be sensitive to the difference in the arrangement of features while quantifying the reliability of a navigation solution. In this work, we propose a consensus-NDT SLAM framework that utilizes measurement consensus metrics to achieve faster convergence and reduce the impact of outliers, feature mismatches, and occlusions on the navigation solution. The measurement consensus metric provides an estimate of the reliability of the navigation solution by a two-tiered quantification approach. First, an NDT approximation is used for the nominal error model on a voxel level. The voxel consensus metric is quantified using a smooth normalized residual squared sum that detects measurement faults. Second, this metric is combined to give the localization consensus metric by calculating the normalized inverse of the Consensus Weighted Dilution of Precision (CWDOP) for all voxels. In this way, the localization consensus metric models the impact of the voxel arrangement on the reliability of the navigation solution. In the LiDAR odometry component of consensus-NDT SLAM, first an intermediate navigation solution is obtained using all features present in the point cloud. After this coarse optimization, low consensus metric voxels are removed and the remaining voxels are used to obtain a fine-tuned navigation solution. The removal of low consensus voxels reduces the impact of outliers, feature mismatches, and occlusions on the obtained navigation solution. Voxel consensus metric thresholding leads to faster convergence and reduced computational load without sacrificing localization accuracy. The mapping component uses the LiDAR odometry solutions as initializations for the optimization and updates the map and LiDAR poses using high consensus voxels. To validate our proposed architecture, a localization and mapping experiment is conducted using real-world data collected on the campus of the University of Illinois at Urbana-Champaign. Results show that consensus-NDT SLAM maintains localization accuracy while providing faster convergence with a lower computational cost. Additional experiments conducted using point clouds from the KITTI Vision Benchmark Suite show that the proposed consensus metric successfully quantifies the goodness of match between the measurement and reference with respect to the navigation solution, irrespective of the algorithm used.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2019-11-26 without embargo terms","The student, Ashwin Kanhere, accepted the attached license on 2019-07-19 at 11:46.","The student, Ashwin Kanhere, submitted this Thesis for approval on 2019-07-19 at 11:52.","This Thesis was approved for publication on 2019-07-19 at 13:35.","DSpace SAF Submission Ingestion Package generated from Vireo submission #14363 on 2019-11-26 at 12:54:11","Made available in DSpace on 2019-11-26T20:35:15Z (GMT). No. of bitstreams: 2 KANHERE-THESIS-2019.pdf: 2494209 bytes, checksum: ba4a6e406fd3cadce50afc5bfb6695de (MD5) LICENSE.txt: 4211 bytes, checksum: 00f29dc20abfd65b5e322bffbff276ed (MD5) Previous issue date: 2019-07-19"],"dc:format":["application/pdf"],"dc:identifier":["http://hdl.handle.net/2142/105713"],"dc:language":["en"],"dc:rights":["Copyright 2019 Ashwin Vivek Kanhere"],"dc:subject":["measurement consensus","LiDAR","SLAM"],"dc:title":["Measurement consensus metric aided lidar localization and mapping"],"dc:type":["text"],"thesis:degree_discipline":["Aerospace Engineering"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["M.S."],"thesis:institution_name":["University of Illinois at Urbana-Champaign"]},"updated_at":"2026-07-22T22:24:44Z"}