This session is an essential opportunity for architects, engineers, contractors, and technology enthusiasts to gain a deep understanding of AI’s influence on the future of BIM and Point Cloud processing. This session will highlight the transformative potential that AI offers, offer a “reality check” on some of the promises of AI, and help the audience stay at the forefront of industry advancements. Presenters will share their work applying AI to BIM and Point Cloud processing in a variety of real-world use cases. Discover how AI’s automation capabilities streamline data extraction, noise reduction, and feature recognition within vast point cloud datasets, revolutionizing efficiency and precision. From construction site optimization to urban planning and beyond, this session showcases AI’s ability to unlock new realms of insight and innovation.
Automated Feature Extraction, Scan-To-Design, From Reality Capture Data in The Cloud
Handling and extracting information from reality capture data (point clouds from lasers, drones, photogrammetry, etc.) has always been an arduous task due its size and complexity. With the advent of the latest hardware technology, the size of point clouds is getting bigger, making these data management issues even harder. In this session, we will walk through a complete reality capture workflow that helps users with feature extraction for AEC design and other applications, extracting directly from the Cloud using Autodesk Construction cloud and AI technologies. This workflow will revolutionize how we store, handle, view, and share point cloud data paving the way for efficient utilization of reality capture data anytime, anywhere in the future.
Ramesh Sridharan, Autodesk
Reality Capture and Machine Learning in Industrial Facilities
The Challenge:
Fireproofing, a cement-like compound that minimizes damage in a fire, is a critical asset in the customer’s refinery infrastructure. It is essential to ensuring a safe working environment; however, it degrades over time leading to an increased risk of substantial infrastructure damage in the event of a fire and worker injury or death from falling pieces if not properly maintained. The customer relied on visual inspections to monitor the condition of their fireproofing. These inspections were time intensive, prone to observational bias, and inconsistent across individual inspectors.
The Solution:
The customer approached Kleinfelder, to capture 360-degree LiDAR scans inside their plants. The customer realized immediate cost savings and safety improvements from applying several different computer vision techniques, synthetic data, and developing a historical dataset of labelled cracks to shift the inspection process from a visual, on-site task to a computer-based one. However, employees still had to manually review and tag every scan, introducing the potential for human oversight, resulting in observation bias, inconsistencies, and some areas of deterioration still being missed. To eliminate the possibility of human error, the client approached AltaML, a leading developer of machine learning-powered solutions. AltaML co-developed a machine learning model that automatically detects fireproofing deficiencies using the scanned plant images.
Mark Franklin, Kleinfelder
Enhanced Scan-to-BIM: Streamlining Workflows with Machine Learning and Human Collaboration
Explore the practical advances in scan-to-BIM automation, where machine learning and human expertise converge for superior results. This presentation focuses on the deployment of deep learning artificial neural networks for detection and semantic segmentation in the modeling process. We delve into how AI interfaces with human modelers to improve efficiency, cost-effectiveness, and quality in building information modeling. Additionally, we consider the broader implications as more of the built environment is digitized due to reduced modeling costs.
Thomas Czerniawski, Integrated Projects
AI for Better Mobile Mapping Data Workflows
Learn about the future of mobile mapping data processing with Cyvl’s data processing and mapping technology. We’ll showcase real-world applications where our AI has been deployed to inventory thousands of miles of roadways for governments and consulting firms, focusing on accurate identification and localization of ROW Assets. Our session will also delve into how we’re revolutionizing Pavement Condition Assessments and the brand-new Cyvl Search Platform for Mobile Mapping Data.
Daniel Pelaez, Cyvl.ai
AR, AI And Digital Twin Tech to Improve Infrastructure Construction And Maintenance Projects
In construction or asset maintenance, decisions are usually made in an environment where some amount of data is available to consider. But all too frequently, that data is not easily accessible for quick decisions, or even worse, the available data is simply bad – inaccurate, incomplete, or inconsistent. In other words, it cannot be used to provide either usable information or actionable insights. Bad data complicates construction and post-construction maintenance, leading to frequent errors, delays, rework and litigation.
Integrated 3D digital twin platforms powered by AI and highly accurate AR help to address this issue. Digital Twin brings all spatial data, including GIS, BIM, point cloud and issue reports, into a unified, always up-to-date view of the job site accessible 24/7 by all stakeholders: engineers, workers, subcontractors and owners. Digital Twin can rely on AI to automate manual tasks, such as issue detection, and engineering-grade augmented reality to complete workflows in the field.
This presentation reviews the AI and AR-enabled Digital Twin technology, real-life use cases, cost-savings and issue-mitigation opportunities, and offers practical deployment advice.
Alec Pestov, vGIS Inc.
