14th April, 2021
Bharath Sankaran

Advancing Construction: Digital Tools For Site Efficiency

Tags: IoT, Robotics, Machine Learning, AI, Industrial Forecasting

The construction industry is extremely complex; organizations are under increasing pressure to be more efficient in project delivery. 

The main concern facing the construction industry is how to bring unproductive sites—sites that are currently resource-intensive, inefficient, and wasteful—into productive use. With the globe’s population rapidly expanding, the construction sector needs to build over 13,000 buildings each day between now and 2050 to support a growing population of 7 billion people living in cities. The need is urgent and as a result, the construction industry is on a mission to modernize.

Over the past decade, we’ve seen the construction technology (ConTech) ecosystem shift toward integrated software platforms that better serve stakeholder needs. Data gathered from construction sites form the backbone of these solutions and facilitate better planning and project execution. While we are beginning to see greater industry-wide technology adoption, construction remains one of the least digitized sectors to date. Currently, 95% of all data captured in construction and engineering goes unused.

In addition, ConTech is heavily fragmented which means that there is a significant opportunity for industry players to create new integration platforms. By embracing digitalization, building projects can benefit from real-time insights gathered from connected devices that result in greater design accuracy and shorter development times.

To gain deeper insights into the ConTech space, we spoke to Bharath Sankaran, Co-founder, and CTO of Scaled Robotics, an organization that is applying robotics and machine learning to build tools that track, analyze, and optimize the construction process.

The Construction Landscape: Motivations for Change

What do individuals not know about the construction industry?
Generally, individuals are unaware of the following three core areas of impact:

  • The immense scale of the construction industry 
    It’s a global $10 trillion industry but it’s extremely wasteful and inefficient. The current set of outdated tools and processes is causing significant losses which forces industry insiders to rethink the execution of construction projects.
  • Its impact on the environment
    The construction industry is one of the largest contributors to carbon emissions with operational processes contributing close to 28% of the global carbon footprint and upfront emissions contributing to approximately 11%. Nearly a third of the material used in every construction project ends up in a landfill.
  • The industry is ripe for innovation
    It’s one of the most outdated industries in terms of onsite tools and processes. This culminates in the industry being extremely wasteful and inefficient.

How is Scaled Robotics addressing issues of waste and inefficiency in the construction industry?
The combination of outdated tools and processes prevents construction site managers and other stakeholders from efficiently tracking progress and measuring quality on their construction projects. This leads to delays and budget overruns caused by rework. At Scaled Robotics, we are developing tools using AI and ML that can measure, track, and analyze construction processes to enable efficient process control.

Why should organizations digitize the construction workflow?
By digitizing the construction process, organizations can use modern workflows like laser scanning to accurately measure the construction site and then compare these measurements to their digital counterparts. This enables stakeholders to have a more accurate overview of the current state of the construction project. Now more than ever, it’s essential that all stakeholders have access to the same information as a central source of truth. The digitization of information also facilitates remote collaboration and coordination which has become even more essential post-COVID.

How is digital buy-in secured from construction organizations using outdated tools and processes?
A classic way to address this problem is to demonstrate the ROI (return on investment) to potential clients and decision-makers. This can be accomplished in multiple ways, however, often the simplest solution is to present case studies from previous projects with new clients. Additional buy-in methods include reducing friction and barriers to adoption. Friction can be reduced by offering potential clients a free trial or subsidized proof of concept (PoC) to test out the technology and see how it can save them both time and money. At Scaled Robotics, in cases where we identify a barrier to entry, we ensure that our technology seamlessly integrates with a client’s existing workflow. Our web-based software is an example of where our clients in the construction sector require little-to-no IT infrastructure to integrate the solution. All that’s needed is a computer and access to the internet which is ubiquitous these days.

If more organizations were to digitize construction, what impact would this have on job security?
Automation doesn’t erase jobs. Instead, it evolves them to become highly skilled and highly paid. For example, when we did away with horse carriages we didn’t do away with drivers. Their jobs became more skillful and rewarding as they evolved to operate more complex machinery like cars and buses. Similar to other industries, the construction sector will see an occupation evolution in the right direction. Although, there’s currently a severe labor shortage in construction. Today’s generation doesn’t find these jobs as interesting as they once were. Technologically speaking, jobs in construction are stuck in a previous century and few want to do the same jobs that their parents or grandparents did. By adopting new technologies, construction jobs will become more sought after for their skill set requirements and high-paying salaries. This will pave the way for a new generation of workers who view construction as a potential career path for growth and innovation opportunities.

Understanding Smart Construction Technologies

What is BIM and its enabling technologies?
Building information modeling (BIM) is considered to be one of the greatest value generators in the ConTech space with 82% of BIM users reporting a positive ROI. By definition, BIM is a digital representation of the physical and functional characteristics of a building. This representation contains precise geometric and semantic information regarding the construction site which is required for the planning, design, construction, operation, and maintenance of physical infrastructures. BIM is enabled by a combination of improved 2D and 3D CAD software along with detailed 4D modeling. Having a centralized source of truth (a BIM file) for infrastructure has improved coordination in digital design, planning and scheduling, AI-based predictive analysis, construction project and documentation management, quality control, and infrastructure maintenance, just to name a few.

How is BIM facilitating automation in the construction industry?
45% of construction professionals report spending more time than expected on non-optimal activities. One of the main reasons for budget overruns and delays in construction stems from the fact that there is no easy and reliable way to compare what has been built to what was originally designed. This verification process is called redlining. The current redlining process is manual, error-prone, and requires days, if not weeks, of work. This flawed process contributes to incorrect progress reports. Automating verification and progress monitoring helps address issues of speed and accuracy introduced by manual processes. Comparing the BIM model to onsite data collected using a reality capture device ensures accurate quality and progress information in time frames magnitudes faster than traditional methods. For example, at Scaled Robotics, our ML and AI algorithms automatically compare laser scans of the construction site to the BIM model and provide precise quality and progress information. This, along with other data products extracted from the same data source, is presented to construction stakeholders to make data-driven decisions that keep projects within budget and on schedule. Not only is this analysis fully automated and precise, thereby eliminating all sources of human error; this information is searchable and localizable with issues ordered by risk and importance.

At present, which industry bottlenecks require additional improvement?
Construction is a complex and non-linear process involving multiple subprocesses (and components) that interact with each other in a myriad of ways. Technological advances can address specific problems either at the component level or at the process level. Instead of building solutions that address each of the individual components within the overall process, it would be more prudent to look at the system as a whole and implement technology that can have an impact across the board. One common way to optimize a complex non-linear process is to centralize the process — all parts see the same information. This can be achieved through the implementation of BIM and BIM-based digitized workflows. The shared BIM would continuously be updated, thereby enhancing planning and coordination. Another major bottleneck preventing construction from evolving is the current industry mindset. A resistance to change is common in most traditional sectors that are labor-centric because labor is used to compensate for failures in data management. The construction industry needs to transition from being labor-centric to data-centric. Closing this information gap would have a tremendous impact and enable the industry to rapidly advance while simultaneously reducing waste and inefficiency. Unsurprisingly this loop closure can only be achieved by adopting BIM and BIM-based digitized workflows.

Why should ConTech organizations leverage tools to reduce waste and inefficiencies?
The need to track and analyze a particular process stems from the need to optimize that specific construction process. This emerges from a simple philosophy, “You cannot manage what you cannot measure”. In technical literature, this is called process control and it’s comprised of three key steps:

  • Measuring a process so that it can be compared to a reference process/model
  • Comparing the measurement to the reference model/process and determining how the actual process has deviated from the design model
  • Utilizing this difference (deviation) between the two processes to correct the measured process, such that it more closely resembles the design model (or vice versa)

This is analogous to what we do when driving on a freeway and staying within the speed limit — a typical example of process control. Imagine you’re driving on an empty highway. The objectives of this driving process are: a) to stay on the road; b) stay within the speed limit; c) arrive at the destination as fast as possible while remaining within the speed limit and; d) reduce fuel consumption which is equivalent to minimizing changes in velocity. The driving process is measured by the speedometer reading. Based on this measurement, the corrective action is the pressure we apply to the accelerator pedal. When other factors are involved in the driving process, like navigating traffic, additional constraints are considered (e.g. having smooth accelerations and decelerations). This would also ensure that the first four objectives are met.

The concept of process control has been widely applied in many industries. In construction, process control works as follows. The reference process/model is available through the BIM. A measurable proxy for the evolution of the construction process is the physical building itself which can be measured using reality capture hardware. With this captured data, quality control tools allow site managers to automatically compare and analyze the deviations between the building and the BIM. This allows them to take the right corrective action, thereby closing the control loop. The faster this analysis is performed, and the quicker the corrective actions are executed, the more efficient and lean the construction process becomes.

What onsite data is being collected and how does it influence corresponding offsite processes?
There are three primary sources of data on a construction site: 1) data about material: (e.g. the physical building structure and onsite material/inventory); 2) data about personnel: (e.g. movement, space utilization, safety compliance, and access authorization) and lastly; 3) data about equipment: (e.g. inventory, equipment usage, and equipment maintenance). Other sources of data include information about the weather and environment, however, these are not widely captured or exploited as of today. This data can influence offsite processes such as quality control, progress monitoring, planning and scheduling, offsite logistics, offsite inventory management and procurement, invoicing and billing, budgeting and forecasting, insurance, health and safety compliance, and litigation.

How are ML and robotics currently being used in construction?
Both ML and robotics play an integral role in construction and should be analyzed separately according to their own merits.

1. Machine Learning Capabilities:
Any process that involves data can be augmented with ML models for efficient and accurate predictions, decision making, analysis, and optimization.

With onsite processes, ML is used for quality control and progress monitoring in construction. It’s also used for material and personnel management (e.g. processes like onsite warehousing, onsite inventory, worker health and safety management, and onsite scheduling, etc.). In addition, ML can be used for offsite construction processes in the design phase, the planning phase, and the execution phase. In the design phase, for instance, ML is used in the generative modeling of building structures. In the planning and execution phase, ML is used to predict optimal schedules based on the current state of the construction project and its associated constraints. Similarly, ML can be used to optimize offsite logistics, budgeting, resource and document management, and supply chain management, etc. 

2. Robotic Capabilities:
While the availability of data acts as the precondition to implementing ML, the precondition for robotics is that the target process should involve either physical movement or physical manipulation. Onsite robotics augment several construction processes like excavation, 3D printing of structures, reality capture for surveying, onsite installation, onsite material transport, and onsite demolition, etc. On the other hand, offsite robotics has had the most traction in prefabrication and warehouse logistics.

Are implementations of robotics and AI improving sustainability in the construction industry?
As one of the most polluting industries in the world, any level of optimization would have a tremendous impact on sustainability. The introduction of robotics and AI in construction reduces waste and increases efficiency. Processes like quality control, pre-fabrication, 3D printing, and robotic installation are all focused on reducing material waste and rework which results in a positive environmental and economic impact. Integrating AI into the construction process could increase industry profits by 71% by 2035.

Are we at a stage where robotics needs to factor in AGI (artificial general intelligence)?
AGI is too far away to be a practical concern for robotics in construction (or any other industry for that matter). Since a construction site is a tightly controlled environment with many ground truth models, the need for AGI is not readily obvious. Currently, an immense amount of value can be extracted by introducing simple automation and AI which is already prevalent in other industries. In general, the discussion about AGI systems in a construction environment might be great for creating hype and publicity in a sector that is starved for skilled labor but the practical implications are limited. There’s no real value in thinking about AGI in the construction sector for at least the next 15 - 20 years.

So, what will the ConTech industry look like in the next 5-10 years?
Though on the surface this seems like an easy question to answer, in reality, it’s extremely difficult. The most probable evolution would be towards more digitization resulting in better planning and coordination which utilizes (new) digitized workflows. Pre-fabrication and offsite processes that allow for better digitization of onsite workflows are another area of evolution. Outside of the information management realm, onsite processes will see advances based on offsite innovations in both materials and methods. This would result in creating new construction jobs requiring entirely new skill sets that were previously considered irrelevant to the construction sector. Such skills would include material science, artificial intelligence, machine learning, optimization, manufacturing process control, and lean management, just to name a few.

Construction Technology in the Machine Economy

How would you define the Machine Economy in your own words?
The Machine Economy enables the autonomous interaction of economically independent machines where each device in the network informs the other about their current state. Each device collectively optimizes a common objective while satisfying individual needs. In construction, this would include the onsite deployment of IoT sensors on all equipment and material. The exchange of information between devices helps inform and optimize onsite processes. When a gateway between this internal network and external nodes is created, it significantly optimizes offsite processes. 

If we were to imagine a motivating moonshot example in the construction industry, it would look something like this: An onsite surveying robot informs a 3D printing robot that material needs to be printed at a particular location. This 3D printing robot informs the material handling system about the amount of material it will require which in turn triggers the warehouse IoT sensors to notify logistics robots to procure and deliver material to the appropriate location. Information from the IoT devices is automatically relayed to an ML-based scheduling system that updates the construction plan and triggers a new forecast of the estimated budget and schedule.

The Next Chapter in Construction Technology

The mandate for change and technology adoption in construction has never been stronger. Within 10 years, full-scale digitization could lead to savings between $0.7-1.2 trillion (13- 21%) in the design and engineering, and construction phases and $0.3-0.5 trillion (10-17%) in the operations phase. 

It’s unlikely that one particular technology will disrupt construction on its own. Rather, it’s more likely that a confluence of technologies will be used in disruptive ways. This will solve many current industry bottlenecks while simultaneously bringing future contractors, engineers, and innovators to the space.With a wealth of construction technologies already in production and many more on the horizon, the future of ConTech is now.

Special thanks to Bharath Sankaran from Scaled Robotics for his valuable input on the topic.

About the author

Bharath Sankaran, Co-founder and CTO at Scaled Robotics
Editing by Megan Doyle, Business Content Specialist at Next Big Thing AG