Thursday, April 19, 2018

Project Virga – Uses Machine Learning and A.I. to Create Parametric System Families

From the BuildingSP website:

Your Firm's DNA 

The architecture, engineering, and construction (AEC) industry is often described as "fragmented." Nothing typifies this description better than our past project CAD files. These files, which were created with much care and at significant cost, are haphazardly stored on hard drives throughout your enterprise.

As a whole, they are your firm's DNA - they describe what you've delivered to projects. (When graphically displayed, it literally looks like DNA.)

We want to help you unlock your intellectual property.

Data Before Automation 

There are many terms for advanced automation in modern computing. In order to leverage any form of advanced automation - Artificial Intelligence, Machine Learning, or Deep Learning - it's important to remember that there is a required element that determines automation performance, scalability, and quality.

All advanced automation needs data.

You Keep Your Data! 

Your data is your data. We will not use your data with other clients or in our products. We believe your past projects are a valuable asset and it will be protected. Your firm's data is your intellectual property.

How Important is Automation? 

How important is advanced automation? If you're a leader of an AEC firm, you should know that advanced automation is either a opportunity for your firm or a threat to your core business. People are the most valuable asset in the AEC industry and Artificial Intelligence, Machine Learning, and Deep Learning are augmenting the skills of people in business all around us. This has yet to occur in AEC at a large scale.

Will your firm be first?

About Project Virga 

Project Virga uses automation to locate, aggregate, and refactor data from your firm's past projects. Your project info, which describes your past deliverables, is then hosted on the cloud in a special form of transactional, enterprise database. Using this data, we perform pattern matching, relationship analysis, and statistical methods to create revolutionary forms of parametric templates for building information modeling (BIM). These templates are flexible, parametric system families (not part families) that could only be created with high quality data, automation, and computing.

This is the application of Machine Learning and Artificial Intelligence in the AEC industry.

Mechanical Room Configurators 

The first focus for Project Virga is the parametric configuration of mechanical rooms, defined using a system programming language called UML. The UML definition of the mechanical room, along with parameters for the systems in the room, will drive the modeling of mechanical rooms while being informed by past project data. The opportunity is to quickly configure mechanical rooms using automation and machine learning, with output to Revit, Fabrication and VR.

Project Virga Details

Want more details? Here's information about how BuildingSP is approaching Machine Learning and Artificial Intelligence in AEC:

  • Initial Focus on Mechanical and Plumbing Contractors: We're focusing on mechanical and plumbing subcontractors during our initial efforts. This is because these contractors have the highest level of development in their models, require a large amount of labor hours to model their work, and have traditionally been technology leaders in AEC. We will expand our scope over time. 
  • Configuration of Mechanical Rooms: Our specific focus is on the configuration of mechanical rooms. Mechanical rooms are important. They are very congested volumes, require room for maintenance, and need a lot of attention during the modeling process. Mechanical rooms need to be high-functioning areas. We're focusing our automation efforts on mechanical rooms so we can configure them using automation, provide options for review by project stakeholders, and allow flexibility over the course of construction planning. 
  • Only Accepting 20 Clients: We've decided to work with only 20 firms who want to be on the forefront of Machine Learning and Artificial Intelligence. We're limiting the number of firms so we can focus our efforts and provide a competitive advantage to our clients. 
  • It Starts with Consulting: The initial engagement with each client will be a consulting agreement, where we work with firms to aggregate and refactor their data, providing a report on our results and providing data access for our clients. The data approach opens up additional avenues for technology innovation. For example, our cloud-based transactional database could be used for the revision management of spool drawings. 
  • Your Intellectual Property, Not Ours: Your data is your data. We will not use your data with other clients or in outside research. We believe your past projects are a firm asset and we will work with companies to protect their data. 
  • File Type and Project History: Participating firms must have enough project history to provide a training set for the next step in the process, which is automation of systems. Therefore, firms must have over 10 years of projects in Autodesk Fabrication.
  • Progressive Approach to Automation: The overall goal is to create revolutionary, AI-driven automation for firms. To do this, we'll incrementally automation individual systems for firms and grow the overall automation capabilities in a progressive way. This also facilitates reasonable contracting engagements.

Still Interested?

If you're interested in more information, reach out to Brett Young at

What's Up with the Name?

"Virga" is a cloud phenomenon. Virga is rain that evaporates before it hits the ground, retaining moisture in the atmosphere. It often occurs during the summertime in the late afternoon, creating a stunning natural display as the sunlight backlights the evaporating rain. More information is available at Wikipedia.

Virga occuring in Australia. Image from Wikipedia:

Copyright © All rights reserved. BuildingSP, Inc. 2017

There's more information available on the BuildingSP website.

No comments: