Automated Pier Damage and Reconnaissance Tool

Pier assets and wharf infrastructure have a critical role in the US Naval strategies and mission capabilities. Expeditionary inspections, assessments, and repair planning requires trained operators, engineers, and a prohibitive number of man hours. The US Navy currently collects three-dimensional (3D) point cloud data on the structures using subsea acoustic multibeam systems (MBES) and terrestrial laser scanning (LIDAR) from manned surface vessels. The 3D data provides a rich point cloud defining the shape and bulk conditions of the structures. Recent advances in Machine Learning (ML) models provides a new opportunity to leverage these data for an automated pier battle damage assessment tool, significantly reducing the time required to assess and make volumetric repair estimates.   Makai proposes an offline expeditionary post-processing tool that uses the strength of neural networks to identify and segment point cloud data into known components of the pier and wharves, and state of the art shape-fitting algorithms to automatically generate “like-new” reconstructions, and then estimate statistical deformations and gaps between the point clouds and reconstructions; providing tabulated and graphical analysis of pier/wharf damage and repair requirements. Critical software modules include data processing, neural network classification, data clustering, primitive shape fitting, deformation and damage analysis, and graphical user interfaces and tabular data outputs. The proposed approach provides an achievable solution to this problem, focusing on systems that can work off the network, are well suited for future commercialization, provide computational efficiency, and perform robustly through proper training of the ML algorithms and strategic use augmented reality GUI design.