Autodesk recently unveiled an experimental, AI-powered 3D object generator called Project Bernini, while NVIDIA announced its working with Google parent company Alphabet's Intrinsic and Isaac robotics technologies to advance autonomous robotic manipulation using its own AI solutions.

Construction management platform Buildots, on May 15, added a new Delay Forecast feature to its construction project management platform. The AI-powered tool predicts project delay risks and alerts building teams about pace issues so they can act proactively to avoid potential delays.

Delay Forecast

Image courtesy Buildots

Looking Ahead

Delay Forecast is available as an update to Buildots' platform and integrates into the project management dashboard used to manage the critical path on projects. Buildots platform uses regular site imagery and reality capture to estimate work progress, noting materials installed and comparing project progress in real time. Trained on similar project data, this new tool tracks project elements such as exact quantities installed and other inputs. It then recommends pace corrections to mitigate and prevent delays.

 "It knows how many quantities are needed in total, then assuming the pace continues this way, you will finish this many days late," says Aviv Leibovici, co-founder and chief product officer at Buildots.

The feature was developed after research performed by Buildots found that 62% of construction site activities consistently run slower than planned, and that 25% of activities run at half their planned pace. It's been beta tested in Europe and Israel. Leibovici said that the tool is intended to help project managers and superintendents who are inundated with tasks to keep projects on their critical path even as construction activities that may not look outwardly like they're behind schedule are threatening to create delays.

Generative AI

Image Courtesy Autodesk

More Generative AI

Autodesk unveiled its new research project, Project Bernini, an AI-powered 3D object generator. Named after 17th-Century Italian sculptor and architect Gian Lorenzo Bernini, the tool's algorithm can take a 2D image or even text inputs to generate a detailed 3D model. 

Autodesk claims that Bernini can differentiate between surfaces, shadows and openings on a 2D image, avoiding the kinds of perspective errors that plague many 3D model generators.

"Embodied in our software is that the topology really matters, the rigorous sort of definition of the surface and its integrity matters," says Daron Green, chief scientist at Autodesk. "We really care about the topology."

Many architecture firms are developing their own AI-empowered 3D object generators for tasks like building parts of a project or for conceptual design. While both Green and Hooman Shayani, senior principal AI research scientist at Autodesk, stressed that Project Bernini is experimental, both said it could someday help designers and contractors generate the 3D objects they need for projects. Autodesk said that, if trained on buildingsits models could someday produce geometrically correct designs.

 "There's a ton of value in the fact that they're actually able to create real geometry that's watertight and intelligent," says Autodesk Executive Vice President for AEC Amy Bunszel.

NVIDIA Partners With Google on Robotics

NVIDIA announced May 7 at the Automate trade show in Chicago that software and AI robotics company Intrinsic is adopting the NVIDIA Isaac Manipulator platform to enhance industrial automation with AI. Intrinsic is part of Alphabet, the parent company of Google.

NVIDIA unveiled the Isaac Manipulator at its GTC trade show in March. Isaac Manipulator is a collection of foundation models and modular GPU-accelerated libraries that help industrial automation companies build scalable and repeatable workflows for dynamic manipulation tasks by accelerating AI model training and task reprogramming.  While not an AEC technology specifically, Isaac-enabled workflows from Alphabet and Intrinsic could help suppliers automate building material manufacturing and even inform prefabrication for construction.

Gerard Andrews, NVIDIA's director of product marketing for robotics, says the chip manufacturer is committed to development of Edge Computing systems which act as the robotic brains of AI-enabled manufacturing robotics, the training computers that train the AI algorithms in the cloud or on premise and through NVIDIA's Omniverse simulation platform.