Commercial construction sites have a wealth of data available in the physical world.
Workers flow between locations, equipment is moved, materials are delivered and installed, and work occurs as planned or deviates from the expected schedule. The abundance of information that arises from operations is not the issue. The challenge is making that physical world knowledge useful to systems that can process it, and extract value.
That is where the combination of AI and IoT becomes interesting to the construction industry
Instead of thinking of IoT and AI as two separate technologies, consider that AIoT represents physical world sensing technologies, data processing and analysis, and actionable intelligence.
The overall architecture can be broken down in a high-level view as:
Physical Assets -> Sensors -> Connectivity -> Data Pipeline -> AI/ML -> Intelligence -> Action
Each element in this list represents a stack within the overall architecture.
1. The Physical Layer
The physical side of the equation refers to the objects and activities on the jobsite.
This could be:
Construction equipment
Tools and assets
Materials
Workers
Vehicles
Jobsite access points
Environmental conditions
and more.
Sensors and identification technologies can turn activities that occur in the physical world into machine readable information.
Depending upon the use case, technologies such as RFID, BLE, UWB, GPS, LoRaWAN and telematics can be used to sense and identify activity at the jobsite. Different technologies can provide varying degrees and types of information, from identification of an object or person, to locational information.
The choice of which technologies to use is driven by the operational need.
2. Connectivity and Data Collection
The information provided by sensors is only useful if it can get to the systems that will process and analyze it.
This is where the connectivity layer comes into play.
Jobsites and their environments present unique challenges due to structures, layout, equipment and other factors that may impact signal strength or reception. Information may flow through gateways or other mechanisms before getting to the overall data pipeline.
This is also where discussion about data quality becomes relevant.
Duplicates, dropped signals, inaccurate location information, disconnected devices or poor identification can degrade the quality of the dataset being used downstream.
3. Building a Usable Data Layer
At some point, information coming from the physical world needs to become something a system can understand.
A construction AIoT architecture may see a series of events being captured in some form like this:
asset_id: 1042
event: location_update
zone: Level_03
timestamp: 10:42:18
source: UWB
Another message might originate from a piece of equipment, or a record that a material has been moved. Events coming into the system are often limited in scope or meaning.
The information gains value when it can be related to the context of the work. That might appear as:
Asset > Project > Location > Activity > Time > Status
By putting a physical world event into context, it becomes something that can be fed into analytics systems.
4. Where AI Fits
IoT is good at answering what is happening.
AI can help answer what does the data mean.
Machine-learning and analytics can provide an array of capabilities, ranging from pattern recognition, anomaly detection, predictive analytics, and helping teams make sense of what is going on.
For construction, that could mean analyzing information around utilization of assets, workers, materials or other factors. The AI layer depends upon the intelligence coming from the data layer, and is limited if the input information is flawed or incomplete.
5. From Tracking to Intelligence
Let's take the example of equipment tracking. A basic system may identify where equipment is located. A more sophisticated system could identify history of where equipment has been. An analytics layer might allow a team to understand how equipment is being used, and if there are patterns that should be investigated further.
Where is equipment X? -> Where has equipment X been? -> Why is equipment X being used this way?
This is one area where AIoT can differ from traditional asset tracking approaches.
6. Connecting With Existing Construction Systems
One area that is often overlooked or not integrated is existing systems.
Many construction companies utilize systems around scheduling, BIM, project controls, drawings, procurement, workforce and more. Information from these systems can be used in combination with AIoT data to help provide context and enable richer analytics.
The potential exists to link BIM + IoT + Project Data + AI/ML to create new opportunities, instead of having isolated data sources.
7. Practical Implementation Considerations
Instead of looking at technologies, organizations interested in construction AIoT should start at the problem to be solved.
Some considerations might include:
What physical world information is not readily available?
What assets or activities need more information?
How accurate do I need location or identification information to be?
How will I validate sensor data?
Should processing occur at the edge, in the cloud or both?
How does this data layer connect to the rest of the organization?
Who will consume the information?
What type of decision does the information need to support?
These types of questions can help create the architecture before looking at specific technologies to use.
For an example of how AIoT concepts can be applied specifically to commercial construction, CommCon AI provides more information.
The Bigger Picture
The biggest benefit of AIoT in construction is connecting the physical and digital worlds.
Sensors and connected devices of various types can capture what is going on at the jobsite, data pipelines can help organize that information, and AI/ML can help produce meaningful and actionable intelligence.
The ultimate value is when that intelligence makes it to the hands of the people and processes that can use it.
The longer term value proposition is not merely sensors, but to create a pipeline of value.
Physical World -> Data -> Contextual Meaning -> Intelligence -> Action
That is what can help transform raw jobsite events into information that can be used by connected construction systems.
Top comments (0)