No abstract

This paper presents algorithms for fast segmentation of 3D point clouds and subsequent classification of the obtained 3D segments. The method jointly determines the ground surface and segments individual objects in 3D, including overhanging structures. When compared to six other terrain modelling techniques, this approach has minimal error between the sensed data and the representation; and is fast (processing a Velodyne scan in approximately 2 seconds). Applications include improved alignment of successive scans by enabling operations in sections (Velodyne scans are aligned 7% sharper compared to an approach using raw points) and more informed decision-making (paths move around overhangs). The use of segmentation to aid classification through 3D features, such as the Spin Image or the Spherical Harmonic Descriptor, is discussed and experimentally compared. Moreover, the segmentation facilitates a novel approach to 3D classification that bypasses feature extraction and directly compares 3D shapes via the ICP algorithm. This technique is shown to achieve accuracy on par with the best feature based classifier (92.1%) while being significantly faster and allowing a clearer understanding of the classifier's behaviour.

Abstract-Change detection is important for autonomous perception systems that operate in dynamic environments. Mapping and tracking components commonly handle two ends of the dynamic spectrum: stationarity and rapid motion. This paper presents a fast algorithm for 3D change detection from LIDAR or equivalent optical range sensors, that can operate from arbitrary viewpoints and can detect fast and slow dynamics. Distinct from prior work, the method explicitly detects changes in the world, and suppresses apparent changes in the data due to exploration at frontiers or behind occlusions. Comprehensive experimentation is performed to assess the performance in several application domains. Sample data and source code are provided.

Objective Risk reduction and increased Fabric Maintenance efficiency using Artificial Intelligence and Machine Learning algorithms to analyze full-facility imagery for atmospheric corrosion detection and classification. Following imagery capture and processing, deficiencies are identified, and targeted mitigation strategies are executed at greatly reduced cycle time and cost. Methods, Procedures, Process A pre-mobilization facility scan plan is generated to maximize imagery quality, including high elevation scan positions, to ensure thorough and comprehensive analytics. Data from all scan positions are stitched together in a point cloud and aligned for accuracy relative to each location. Finalized imagery and point clouds are then tagged with unique piping line numbers per design, fixed equipment tags, or unique asset identification. The Machine Learning algorithm is intensely trained with manual ground truth inputs prior to analysis. The algorithm analyzes each pixel throughout the facility and detects, classifies, and reports on all identified corrosion, tagging faults to specific piping or equipment. Results, Observations, Conclusions Atmospheric corrosion is the number one Asset Integrity threat in the Gulf of Mexico. Utilizing this tool, we can have a comprehensive and objective analysis of a facility’s health in a matter of weeks from the time of data collection. Data collection for a large deep-water, spar facility requires approximately 12 days with 8 data scanning personnel. Conventional manual inspections incur higher risk, higher cost, and reporting is much less objective considering the number of inspectors involved and the duration of a full-facility campaign. Finally, all results are published in a user-friendly dashboard that can be filtered by process type, equipment type, corrosion severity, and many other criteria as the user requires. Each fault is associated with the specific equipment identification and the user can navigate to see the imagery of the corrosion in a 3D, photogrammetric environment. Remediation strategies can be collated into work packs for fabric maintenance teams, further Nondestructive Examination (NDE) assessment, or work orders for replacement. Fabric maintenance efficiencies are substantially realized by targeting decks, blocks, or areas with the highest aggregate surface areas of corrosion (on process equipment or structurally, as selected by the user) and concentrating remediation efforts on at-risk equipment. Novel/Additive Information This application of Artifical Intelligence and Machine Learning is a first-in-industry approach to having a comprehensive understanding of facility coating integrity and external corrosion threats. HSE analysis, Risk awareness, and targeted remediation strategies will make the Asset Integrity program more efficient, proactive, and reduce down-time across the Gulf of Mexico related to atmospheric corrosion.