Take a deep dive into four specific use cases of applying artificial intelligence in construction—scheduling and planning, measuring site progress, earth moving and safety.

AI for Construction Scheduling and Planning

Margins in the construction industry are low. According to Rene Morkos, founder of Alice Technologies, this is partially because the industry is stuck in what he calls The Lonely Dot Universe: There's one dot, one option.

In other words, a contractor might propose to construct a building with a bid of $100 million with one year to complete. But how fast is this? How expensive is this? How to evaluate how good an option this is? By comparing to other competing bids? But how good are they?

Thirty years ago, semiconductor designers used computer aided design tools to simulate thousands of design options before committing substantial funds and and time to fabricate designs into silicon. If the tiniest mistake was discovered at test that meant thousands of dollars wasted and at least a month’s delay before being able to potentially generate revenue in production. One mistake and profit margins dwindled.

Where’s the ability to simulate design options in the construction process? Where’s the computer aided design for construction?

Alice Technologies is doing just that. They are applying artificial intelligence to enable contractors to run millions of building simulations for a single project. They claim their customers consistently save 15-17% on construction duration and 10-13% on labor and equipment costs. Their customers accomplish this by being given the option to simulate various building options to come up with the most efficient configuration that meets their needs – before building.

Alice’s customers can simulate, for instance, what would happen to costs and schedule if they added in a crane? What if they added or subtracted a crew? How does fast dry concrete affect cost and schedule? How can weather impact the schedule?

With each press of Alice’s construction planning software button, the tool builds the new configuration and automatically generates a schedule and 4D model.

At the core of Alice’s secret sauce is an AI engine that does all the heavy lifting of simulating and analyzing millions of schedules in just minutes to help a construction team find the optimal way to build.

AI to Measure Actual Site Progress

There is a lot of talk about digital transformation for construction, which includes connecting a digital world to the real world. But no matter how sophisticated a digital model, someone or some thing (in the case of robotics) still has to build the model in the physical world. So how do contractors know if what they are building is actually on schedule according to the model?

One company addressing this challenge is Doxel, which has an AI-based solution that tracks progress on construction projects and provides real time feedback on progress and quality.

To do this, Doxel uses off the shelf hardware to create a camera-equipped robot that autonomously roams the construction site to capture three-dimensional point clouds of the construction projects. Through these scans, Doxel creates basically a digital twin of the job site.

Then the real magic can begin. Doxel has developed a neural network that automatically processes all of the data, cross references this against the BIM and bill of materials to tell project managers how much is owed on the multi-million-dollar project, how close to schedule are they running and if there are any quality errors. Effectively, this gives project managers X-ray vision into how their projects are performing on a week-to-week basis and allows them to react almost immediately to inefficiencies and errors.

AI to Move Earth

There is also a labor shortage in construction. Autonomous cars are being tested for consumer use. Linking the two, Built Robotics brings to market self-driving heavy equipment. To do this, they upgrade off-the-shelf heavy equipment with GPS, sensors, cameras and AI guidance systems to enable them to operate autonomously. Built Robotics upgrades can be installed on standard equipment from any manufacturer.

Built Robotics is being used by the earthmoving industry for infrastructure projects such as building wind farms, gas pipelines and new housing developments. They claim their autonomous heavy equipment have operated for 7,500 hours (equivalent to driving 350,000 miles in a car) without a single safety incident.

AI to Increase Safety on the Job Site

To help identify safety risks on the jobsite, SmartVid’s platform provides safety observations, safety monitoring and predictive analytics while aggregating all sources of photo and video content into one system powered by its AI engine, Vinnie. Vinnie is trained to find indicators of project risk in the areas of safety, productivity and quality. Vinnie's observations create objective dashboards and reports that rank each project based on potential safety risk. Smartvid.io has several pre-built integrations with construction systems such as Autodesk BIM 360, Procore, Oracle Aconex, OxBlue site cameras, StructionSite and Box.

One customer, Shawmut Design and Construction, already had in place a strong set of policies and technology, especially when it came to early risk indicators. The company deployed Procore’s construction management system for field data collection from project teams and safety personnel. The company also deployed ConstructSecure, as a major component to its trade management risk program. In addition, many projects generated visual project data daily through OxBlue site cameras. The company’s safety team then managed observation-based metrics for risk indication.

However, the Shawmut team knew AI could expand its perspective, providing additional risk data and even predicting issues before they happened. Therefore, they deployed Smartvid’s AI engine Vinnie to build dashboards that helped rank projects by potential risk factors. Vinnie uses construction-specific visual AI models that have been trained to identify indicators of risk in photos, video and other project data. For Shawmut’s team, this included automatically tracking work seen at height, housekeeping, standing water and workers missing personal protective equipment such as hard hats, high vis, safety glasses and gloves. These factors are reported weekly by project to show which areas are improving or becoming less safe.

AI is nothing new. In fact, John McCarthy first coined the term in 1956 when he held his first academic conference on the subject. And although it might have taken a few decades to find itself into the job site, AI for construction use cases have become a technology worth consideration by the construction industry.