It’s no secret that construction projects are notorious for their productivity problems. It’s generally just expected that large construction projects will end up way over budget, and way past their deadlines. But, startup Doxel thinks they can dramatically improve the situation using robots and deep learning.
The robot itself is a rover riding on tank tracks, equipped with a high-end LIDAR system. It meanders around a building under construction, and gathers high-quality 3D data and images of the entire building. Doxel then takes that data and feeds it into a deep learning system for processing.
With that data, Doxel can determine if the overall project is on track. It can also find potential problems with the progress made so far, so that they can be corrected before they turn into real problems. Some of these capabilities rely on the project being accurately modeled in BIM (building information modeling)—something that’s still relatively uncommon—so that the deep learning algorithms can find discrepancies. But, the system is smart enough to track a lot of progress, even without that.
In a test study of the construction of a medical building in San Diego, Doxel claims they saw a 38% increase in labor productivity, and ultimately the project came in 11% under budget. On the scale we’re talking about, that’s a massive amount of money, so Doxel is absolutely making some big claims here. But, if they can show consistent results that amount to even a fraction of that, they’re going to be successful.