NVIDIA is moving aggressively to maintain its substantial lead in autonomous vehicles, moving initially from cars and trucks to drones and, most recently, with Komatsu using the Jetson platform, to construction equipment. This comes on top of advancements in 3D printed buildings and suggests that sometime in the 2020s, we may be able to role up to a construction site, set a series of robots in place with minimal monitoring, and come back a few weeks later and have a completed home, office building, shopping center (if we are still building those by then) or office complex.
Let’s talk about how you’d evolve a robotic construction solution.
The Unique Difficulties and Opportunities of Construction
I drove my first Caterpillar (A D2, if memory serves) when I was around five years old and have been involved in construction on and off for much of my life. What lends this to automation is the extensive amount of planning and documentation that goes into this kind of an effort, which must be approved and certified by inspectors. Two things work against this effort. One, the plans have wriggle room and are often incomplete, and two, this is a heavily unionized activity and unions are typically not that happy about losing union construction jobs.
Now one of the interesting things about the current generation of deep learning AI systems is that they can increasingly deal with variances. If you implement central control and digital prototyping, you basically build the structure in simulation first, thus correcting any mistakes or omissions long before you start construction. This is something we likely should have been doing more of today as it would cut dramatically the common causes for large construction cost overruns.
The nice thing about full digital modeling, and NVIDIA did this with its new headquarters building in Silicon Valley (it kind of redefined “open workspace”), is that you can virtually walk through the project and request changes to the design with minimal cost and see the cost outcome almost immediately for your changes. One of the other big reasons for cost overruns is changes that the owner, architect or inspector require after construction starts.
During this phase, a deep learning system could also suggest (and even implement automatically) changes that would make the project safer, easier to defend against attack, have better traffic flow (vehicle or human), and optimize sun thermal loading to reduce heating and cooling costs. All while providing real-time feedback on the increases and decreases in construction cost.
Once the project was modeled, you would than optimize the local and remote construction process based on transport costs, availability of fabricating resources, and the availability and quality of localized labor.
Once fully emulated and optimized, the system would then specify the construction equipment you’d need and the amount of time you’d need it, even anticipating weather impact based on historical models. Given that Mother Nature seldom cooperates, in certain climates, you might also adjust the amount of remote manufacturing to assure the ability for weather to adversely impact the project was minimized. And, increasingly, you’ll be able to 3D print the structure on site.
Once deployed you’d likely have phases where certain kinds of equipment would be swapped in and out based on the phase. Phase one grading, phase two foundations, phase three framing/component installation, etc.
You’d probably want centralized control to manage traffic and loop in a series of drones, both airborne and rolling to assure tolerances, security, and provide an additional level of worker safety. (Though, at some future point, you might not need human workers for anything but final finish work.)
Wrapping Up: We Can Start Now with This Automation
I don’t think we are even talking about the tip of the iceberg when it comes to where we are going with autonomous systems. In a few short years, we may be able to pick a new house or building out of a catalog, walk through it using VR headsets shortly thereafter, and then have a build time measured in days, not weeks or months, based on our level of customization.
Other than hearing the union screams from space, I expect we’ll also be impressed about how much the quality comes up and the costs go down, largely because the things that increase those costs are eliminated before anything is built.
As a side note, you know we can do a lot of this cost elimination today much like NVIDIA did with its new headquarters, and it still amazes me that more individuals and companies haven’t gone this direction to reduce construction cost.
In the end, we are moving toward a time when virtually anything can be built by robots.
Rob Enderle is President and Principal Analyst of the Enderle Group, a forward-looking emerging technology advisory firm. With over 30 years’ experience in emerging technologies, he has provided regional and global companies with guidance in how to better target customer needs; create new business opportunities; anticipate technology changes; select vendors and products; and present their products in the best possible light. Rob covers the technology industry broadly. Before founding the Enderle Group, Rob was the Senior Research Fellow for Forrester Research and the Giga Information Group, and held senior positions at IBM and ROLM. Follow Rob on Twitter @enderle, on Facebook and on Google+