Abstract: Building Information Modeling integrates 3D models with physical and functional characteristics of a building and infrastructure project and has the potential to facilitate the exchange of information between different parties involved in the same project throughout its lifecycle, ranging from design and construction to maintenance and operation, and beyond. However, the fundamental problem -- lack of interoperability (ie. the inability to exchange information between different platforms) due to model inconsistency and missing information -- prevents the exchange of such information. Existing research efforts to address this lack of interoperability have been heavily focused on standardization and semantic modeling. These standard methods do not address the underlying problem and still depend on the computer models involved. In this talk, I will introduce the concept of invariant signatures of architecture, engineering, and construction (AEC) objects, its discovery and application in solving BIM interoperability problem in a radically different approach from the existing efforts that are focused on data schema standardization and/or term-based semantics of AEC objects. Invariant signatures of an AEC object are defined as a set of intrinsic properties (e.g., geometry, location, material) of the object that distinguish it from other objects, and they do not change with software implementation, modeling decisions, and/or language and cultural contexts. An interdisciplinary approach involving geometry theorems, computer algorithms, and material mechanics was employed to explore and quantify these intrinsic properties. The underlying hypothesis is that invariant signatures of an AEC object collectively defined by the Cartesian points-based geometric, relative location and orientation, and material mechanical properties will enable seamless and universal interoperability of building information modeling (BIM) software in various analysis phases from architectural design and preliminary structural design to detailed structural analysis and construction cost estimation. Successes in applying invariant signatures to different BIM interoperability scenarios will be introduced.

Abstract: This presentation overviews the transformation of civil and environmental engineering education and relevant research efforts at Carnegie Mellon University. Decades of computing research in civil and environmental engineers show the importance of leveraging computational tools in supporting civil and environmental engineers in data-driven simulation, learning, and decision-making. These computational tools are forming a new vision of digital twins in civil and environmental engineering and the underlying convergence of human understanding, machine learning, and Artificial Intelligence in enabling the resilience of civil systems in climate changes. Undergraduate and graduate education and research are under systematic updates at Carnegie Mellon to enable a unified learning and decision-making framework to support learning experiences while fostering a data culture for future civil and environmental engineering projects. The presentation will highlight the threads of sensing and computing in the curriculum at both undergraduate levels, emphasizing the emerging online certificate of “AI Engineering - Digital Twins and Data Analytics.”

Abstract: Numerous infrastructure (bridges, dams, highways, tunnels, etc.) in USA and around the world are reaching their life expectancy, and thus have strong needs for routine inspection and maintenance to ensure sustainability. It is reported that 42% of over 600,000 highway bridges in the National Bridge Inventory (NBI) have exceeded their design life of 50 years, and 42,951 bridges are rated in poor condition and classified as “structurally deficient”. Although normal bridges are inspected biennially (deficient bridges are inspected annually), study has found that the visual inspection results can have a high level of inconsistence in identifying surface flaws. To inspect the structural integrity of bridges, the inspectors also need to detect subsurface defects (i.e., delamination, voids) using NDE instruments such as GPR and impact-echo (IE) device at difficult to access components (i.e., pier, bottom side of the deck). The current practice of manual inspection using hand-held NDE devices by a “spider-man” with rope access, or by using scaffolding or snooper truck has to block traffic, and is expensive, time-consuming, and exposes human inspectors to dangerous situations. This presentation will introduce climbing robots developed over the years at CCNY Robotics Lab that integrate the robot control and vision-based accurate positioning with NDE signal processing to detect both surface flaws and subsurface defects. These robots combine the advantages of aerodynamic attraction and suction to achieve a desirable balance of strong adhesion and high mobility. For example, Rise-Rover with two drive modules can carry heavy payload up to 20 Kg, and GPR-Rover can carry a small GPR antenna for subsurface flaw detection and utility survey on concrete structures. Empowered by the capability of low cost vision-based accurate positioning, the GPR-Rover is able to scan the surface in arbitrary trajectories and tag GPR samples with position information that enables high-resolution 3D GPR imaging. The use of the robotic inspection tool will eliminate the time, hassle and cost to layout grid lines on flat terrain, and make it possible to automatically collect NDE data with minimum human intervention. These robots can reach difficult-to-access areas, take close-up pictures, perform contact-based NDE data collection for further analysis. This presentation will also introduce machine learning algorithms for visual inspection to detect and measure cracks; IE data processing methods that utilizes both learning-based and classical methods to interpret the IE data and reveal subsurface objects; and DNN-based GPR data analysis software to reveal subsurface targets for better visualization. The correlation of findings from multiple NDE sensors (visual, IE and GPR) on board the robot gives a comprehensive evaluation of concrete structures. Field tests demonstrate that the robotic inspection system significantly increases the data collection speed, and make the full bridge inspection faster, safer, and cheaper without affecting traffic flow on roadways.