AI Will Take Construction Robotics from Hype to Reality

Print Friendly, PDF & Email

There is no doubt robots will have a big role in the construction industry, however until these machines can automatically prioritize tasks, project managers will face the same challenges due to inefficiencies, that are amounting to $1.6 trillion a year according to a McKinsey report.

The general awareness AI and deep learning can bring to the table are the first steps to make robotics useful in construction. Building the first steps to unite these technologies, Buildots, an AI startup, attaches 360 cameras onto project managers’ hardhats to collect footage inside the construction site and analyze the data to deliver comprehensible results.

The data is compared to the building’s design plan, providing project managers with a broad view of the design plan, flagging problems that can be then solved early into the project, and tracking progress with greater accuracy. Buildots provide a general awareness that can be used to integrate other technologies like robotics to construction plans without having to spend time and resources directing them.

Buildots platform creates the “construction control room,” by automatically analyzing data captured using hard hat cameras, and comparing it to the designs and schedule. The analysis process is highly accurate and in high resolution, and analyzes every electrical outlet, wall, or window, separately to determine it’s exact state vs. the plans.

Deep learning models are a crucial part of the Buildots platform, and are used throughout the analysis process in multiple flows:

  1. Person Data Removal – clearing people, phone/tablet screens, paper notes, etc
  2. Alignment of 360-degree images – AI-based image stabilization engine
  3. Image quality rating – determining the level of blurriness, darkness and other factors that would make the image less relevant for analysis or human use.
  4. Indoor navigation – highly accurate indoor positioning process that is fully reliant on visuals, based on multiple models that find features, assess room structure and others.
  5. Status classification – realizing the status of every element/task on the construction site using hundreds of different models.

Those models together, accompanied by classic computer vision and machine learning algorithms, do the unbelievable work of transforming random visual data from construction sites, into actionable insights that are correlated with the project’s designs and schedule.

Sign up for the free insideBIGDATA newsletter.

Join us on Twitter: @InsideBigData1 –

Speak Your Mind



  1. This is a fascinating application of AI. I wonder if the same could be done using helmet-mounted LiDAR scanners to product point clouds of the site as it is constructed.