Along the Autonomy Road

Construction industry builds technology momentum.

From technology manufacturers and OEMs to construction firms, large and small, autonomous and semi-autonomous activities are no longer a distant, unrealistic scenario. In fact, if forecasters are even close to accurate, autonomy in construction is closer than we think.

DPR crews operate three different robots on a live jobsite: Hilti’s Jaibot drilling robot, the LightYX BeamerOne‏‏ construction layout tool and the Dusty Robotics FieldPrinter, a BIM-to-field layout tool that automatically prints drawings at 1:1 scale on a construction site floor. Image: DPR Construction.

In a recent study, consulting firm FMI and technology firm Hexagon surveyed more than 1,000 technology leaders from commercial general contracting firms in the U.S., UK and Australia to better understand where and how autonomous technology is used in construction and its role in solving the construction industry’s most pressing productivity and workforce problems.

Some might be surprised that, of those surveyed, 84% are using autonomous technology in their operations. In an industry regularly maligned for its cautionary approach to technology, it’s an interesting finding. A deeper dive into those results shows it’s the automated and, in a growing number of cases, AI-enabled activities within project management workflows making up that high percentage. In the study, actual equipment, aka robotics, are the least deployed thus far. 

But even that has begun to change. The Business Research Company’s Autonomous Construction Equipment Global Market Report 2024 forecasts the autonomous construction equipment market size to grow from $14.05 billion in 2023 to $15.41 billion in 2024 at a compound annual growth rate (CAGR) of 9.7%. 

While truly autonomous equipment is still in the future, the foundational components—advanced (and affordable) sensors, AI, remote-control systems and the means and methods to facilitate real-time data management and insights—are all evolving at a rapid pace. 

Dusty Robotics RPD 35 autonomous pile driving system will begin work at solar sites across the U.S. over the coming year. Image: Built Robotics.

The Robotic Evolution

The most recent IMARC Group report forecasts the global construction robots market size to reach US$504.9 billion by 2032, a growth rate of 14% from 2024-2032. That investment and subsequent development, according to the researchers, is largely attributed to the need for better productivity with fewer workers as well as the industry’s heightened focus on safety and precision, a rising emphasis on sustainable practices and the rapid technological advancements in AI.

Robots are currently available to do everything from laying bricks and tying rebar to driving piles and mapping jobsite progress. 

The Los Altos-based Toyota Research Institute (TRI) is working to improve the dexterity of robots. Last year, TRI announced a breakthrough generative AI approach to behavior learning that’s based on Diffusion Policy that teaches robots new, dexterous skills quickly. TRI put this technique to work on its custom-built dual-arm robot. The group believes this advancement significantly improves robot utility and is a step toward building what TRI calls large behavior models (LBMs) for robots, analogous to the large language models (LLMs) that have recently revolutionized conversational AI.

TRI has already taught robots more than 60 difficult dexterous skills using the new approach, including pouring liquids, using tools, and manipulating deformable objects. The TRI researchers noted that these achievements were realized without writing a single line of new code; the only change was supplying the robot with new data. Building on this success, TRI has set an ambitious target of teaching hundreds of new skills by the end of the year and 1,000 by the end of 2024.

Industry construction leaders are also challenging the status quo. Black & Veatch, Turner Construction, Skanska and many others are pushing technology innovations from the lab and test sites to real-world environments. 

“We see AI as an essential piece for enabling the remote monitoring and control of construction sites from centralized locations, taking over human decision-making roles.”

Dr. Tony T.Y. Yang, Professor, Structural and Earthquake Engineering, University of British Columbia

Commercial self-performing general contractor firm, DPR Construction sees tremendous potential in technology (and robots) to drive productivity. The firm is at the forefront of those innovations, investing time and money into the development and testing of remote controlled, semi-autonomous and autonomous solutions. 

In the near term, the Redwood City, California-based firm is particularly interested in improving site logistics through automated materials and equipment movement. Henning Roedel, the robotics lead for DPR, said, “We view robotics and automation as tools to handle dull, dirty, dangerous work, freeing up craftspeople for skilled tasks. It’s why we are partnering with technology providers to automate processes like site logistics.”

DPR already uses the LightYX Beamer One construction layout tool, which projects 2D or 3D plans onto a construction surface to scale. This checks designs against existing conditions and updates those drawings/models accordingly, using a mobile app. They also rely on the Dusty Robotics automated site layout tool to support multi-trade layouts and the Hilti Jaibot semi-autonomous mobile ceiling-drilling robot to support mechanical, electrical and plumbing installation contractors. 

More automated, autonomous and intelligent applications are on the way. For instance, working closely with Field AI, a robotics engineering firm headquartered in Irvine, California, Roedel and his team have been testing autonomous reality capture on a warehouse project for the last five months. Using Field AI’s AI-powered quadruped robot dog, the project team is looking to not just gather information about the jobsite, but have this robotic system make decisions about where it should walk and the tasks to be completed, with the goal of reducing human interventions over time. 

Roedel said, “We see the value in having a robot that can navigate construction sites and make judgments about what materials/terrain a robot can traverse safely.”

Thus far that pilot project is going well. “We calculate progress by the number of interventions required during a given mission. We’ve gotten down to two interventions, which I see as great progress,” he confirms. Advancements and adjustments on the project will continue through 2024. 

Bobcat’s RogueX2 autonomous track loader is a concept machine used to test various capabilities including remote control, collision warning and avoidance systems and autonomous capabilities. Image: Bobcat

Aerial Advantage

Flying robots—aka drones—are also expanding their reach on construction sites. 

From topographic mapping and land surveys to equipment tracking and earthwork volumes, UAVs are an integral part of construction projects across the country. It’s an area of investment and advancement that DPR’s Henning is especially excited about. He points to a study just published by Oregon State University researchers that found a “swarm” of more than 100 autonomous ground and aerial robots can be supervised by one person. 

Henning continued, “They set it up with objectives and let the autonomous systems make a lot of the decisions. While I don’t see that happening on a construction site, the potential for UAVs to take on roles beyond basic material management and progress tracking is powerful. Imagine if we leverage decades of tooling and support from the manufacturing industry to advance industrialized/prefabricated construction?” 

Public entities that once pushed aside UAV-gathered data are now adapting contracts to request, and in some cases, require it for design, progress verification and even maintenance activities. 

For instance, the Pennsylvania Dept. of Transportation (PennDOT) is using aerial imaging to find ways to alleviate standing water and drainage issues on roads and infrastructure. The agency recently partnered with T3 Global Strategies to gather topographic planimetric mapping data using drones to create the ground surface for engineering analysis.

The Construction Drone Global Market Report 2024 noted that the construction drone market grew from $6.01 billion in 2023 to $6.99 billion in 2024—in part due to infrastructure development and labor shortages. Those numbers are expected to almost double by 2028, thanks to a number of factors including the adoption of 5G. 

Built Robotics’ Ahmed adds, “Real-time connectivity is important for backup, management and data analysis. Even though the robots can work in limited network environments, strong network connectivity unlocks the full potential of automation and AI. Advances in 5G and satellite networks are improving the reliability of connectivity even in remote locations like solar farms.”

Hitachi is focusing its efforts on developing transmission channels to connect sites and remote offices. It is working on proving trials of Solution Linkage Wi-Fi wireless LAN and equipment, which allows a single operator in one remote control chair to work over a regular 5G line and switch between multiple machines, picking the right one for each task.

Using the Bobcat MaxControl remote operation application, operators can operate machines from outside the cab. Future features will include known object avoidance, unknown object avoidance, program navigation, key components of a fully autonomous machine. Image: Bobcat

From Trucks to Trenches

Autonomous machines, though in limited number, have begun to emerge in the industry on test sites, on the job and even on the road. 

The Colorado Department of Transportation (CDOT) is using a federal Strengthening Mobility and Revolutionizing Transportation (SMART) grant to expand the use of an Autonomous Truck Mounted Attenuator/Impact Protection Vehicle, a self-driving truck designed to protect workers in construction zones. CDOT currently has two automated safety trucks in its fleet, which are primarily used during lane-striping projects.

Built Robotics continues to expand its fleet of autonomous construction solutions. Adding to its autonomous trenching excavator that has been out for several years, the firm introduced the RPD 35 autonomous pile driver, a solution ideal for large-scale solar farm construction projects. The autonomous pile driving system will be deployed on solar sites throughout the U.S. in states such as California, Texas or Florida over the coming year. 

As well, Built Robotics acquired Roin Technologies, the developer of robotic concrete finishing solutions, including a shotcrete robot and a concrete troweling robot. The acquisition is helping to power the piling robot from Built. 

Erol Ahmed, director of communications for Built Robotics, said, “We continue to advance our AI and machine learning systems on these machines. We’re tracking technology advancements in areas like sensor fusion, vision systems, and data processing that could further enhance the capabilities and cost-effectiveness of autonomous construction equipment over time. For instance, we’re working on new enhancements to our piling robot around automated quality control using computer vision to help verify pile positioning accuracy.”

Aerial Advantage

Fusing sensors such as GNSS, IMUs, cameras, radar, laser scanners and LiDAR is already improving positioning and perception on construction equipment operating in complex job sites, while building on autonomous capabilities.

For instance, high-speed 3D laser scanning technology is integral to the development of autonomous solutions. In one example, Trimble and Exyn combined the ExynPak SLAM-based mapping solution powered by ExynAI and the Trimble X7 total station solution for reality capture to achieve level four autonomy. 

Similarly, LiDAR provides high-resolution sensing capabilities important for autonomous machines to perceive their environment and for object detection and avoidance. It’s already an integral component of autonomous excavators and pile drivers, mobile mapping systems and UAVs. As LiDAR costs continue to decline, the technology benefits in the construction space will also expand. 

Giri Baleri, director of product management and strategic marketing, OEM GNSS at Trimble, notes, “The cost of LiDAR sensors has declined exponentially in the last 10 years, making them more affordable for construction equipment applications—and a clear enabler for capabilities such as situational awareness and obstacle avoidance.”

But LiDAR and laser scanners have limits in the construction space. One of the biggest challenges to using LiDAR or vision-based camera systems is they operate in the visible or near-visible light spectrum, wavelengths that have difficulty seeing through dirt and dust kicked up during a construction project. 

Radar could be the answer. One manufacturer is building on existing radar capabilities to develop its emerging autonomous capabilities.

The Look Inside a Fully Autonomous Jobsite

The University of British Columbia (UBC) Smart Structures Lab in Vancouver is working to combine autonomous construction equipment, AI and drones. Last year, the team deployed smart construction robots on a test site on Mitchell Island in Richmond, British Columbia. First, aerial drones fitted captured details on the ground that were then used to create a digital twin. Then, AI-equipped autonomous cranes and forklifts used this information to move construction materials such as beams and columns around the actual site, navigating around obstacles.

Dr. Tony T.Y. Yang, professor of Structural and Earthquake Engineering at UBC, confirmed that the smart construction robots are able to recognize objects, performing detailed scans of structural components for quality assurance. 

“Not only can these robots precisely place objects on site and check that placement against a computer model, but they can make autonomous decisions such as navigating around obstacles or instantly stopping work to protect a worker who is in danger,” he says.

For instance, if a robot is asked to build a wall and senses an unexpected obstacle, it can decide how to navigate around the obstacle and complete its task. The UBC Smart Structures Lab team is working with several construction firms to deploy smart robots on commercial projects.

He continues, “Our research aims to robotize the traditional crane and develop automatic construction technologies to replace the traditional construction methods. We see AI as an essential piece for enabling the remote monitoring and control of construction sites from centralized locations, taking over human decision-making roles.”

So far, a number of deep learning (DL) algorithms have been implemented to identify and localize construction materials on the construction site. For robotized crane motion planning, reinforcement learning (RL) algorithm is used to conduct 3D lifting path planning and transport the objects with collision avoidance. 

The Smart Structures Lab team is testing the autonomous excavator technology at construction sites with industrial partners to validate it can accurately dig based on computer models. Yang is also developing a fully autonomous excavator, which will use drawings to dig out the foundation of buildings without human aid.

A key area of development that must advance to make this scenario possible is connectivity. Yang points to the need for reliable data transmission, e.g., 5G, to make autonomy truly work on a construction site. He adds, “Ensuring secure and reliable data transmission over long distances for remote control of equipment to avoid delays or bottlenecks is essential and we’re working to address this challenge.” 

To facilitate that development, UBC has teamed with Rogers Communications Inc., a Canadian telecommunications company, to complete the development of remote operation capabilities for construction equipment…an effort that could see significant advancement in 2024. UBC has also partnered with Bird Construction, a large construction company in Canada, to test autonomous robots performing tasks on a jobsite with improved 5G connectivity, setting the stage for a future entirely autonomous site.

The University of British Columbia Smart Structures Lab is developing a number of smart construction robots and systems to automate the construction site, including an autonomously operated crane that can precisely place objects on site and make autonomous decisions such as navigating around obstacles or instantly stopping work to protect a worker who is in danger. Image: University of British Columbia.

A Radar Reaction

This year, global compact equipment manufacturer Doosan Bobcat (Bobcat) is set to commercialize a collision warning and avoidance system for its compact equipment, both manned and unmanned, using imaging radar technology. 

“We invested in a startup several years ago because these sensors are much better suited to construction work where there is dust, rain, snow, etc., and they’re affordable,” said Joel Honeyman, vice president of Global Innovation at Bobcat. “We’re trying to make investments that are cost effective. You can’t charge 30% of the price of a machine and think you’re going to commercialize it. It just doesn’t make sense.”

The test bed for all of Bobcat’s advancements, whether new sensors or software applications, is the RogueX2 autonomous loader concept machine. The RogueX2 is an all-electric and autonomous concept machine that produces zero emissions, and features a lithium-ion battery, electric drive system and electric actuated lift and tilt kinematics with no hydraulics. It integrates technologies like remote control, collision warning and avoidance system sensors, and potential for autonomous control that the company is developing for construction equipment.

Honeyman said, “This concept vehicle is a playground for us to test new autonomous and connectivity solutions and to understand customer needs and reactions. While it may not go to full production, aspects of it inform future product designs.”

The RogueX2 is also a canvas for future unmanned machine design. Honeyman adds, “Without an operator station on the loader, we can change the machine geometry to expand capabilities and increase reach for tasks like lifting and dumping. Then a vehicle like this takes on whole new dimensions in the construction space.” 

The next step in Bobcat’s development testing is a control box for advanced 
automated machine communication such as GNSS and cellular connectivity, so the machine can semi-autonomously move from point A to point B, or perform tasks such as move the lift arm, run an attachment, etc. Future Bobcat development, like many others, will include leveraging AI to refine the many use cases for the application of this technology.

Elevating Innovation

One area of advancement that is still in the earliest stages is 4D imaging radar for construction applications. Traditional radar systems focus on horizontal angle, distance and velocity—whereas a 4D radar adds elevation. Like with autonomous vehicles on the road, 4D imaging radar is a Level 4 and 5 enabler. 

These solutions are already beginning to emerge. German-based AI startup sensmore is at the forefront of leveraging novel 4D imaging radars. The company’s unique AI powered products are able to build-up an unprecedented understanding of the harsh environments where heavy machinery operate, delivering driver assistance and autonomous capabilities to industries such as mining, construction and agriculture. Its technology is being deployed in surface, hybrid, and underground environments and operates even in the most severe conditions where today’s systems fail.

Cameron Clark, earthmoving industry director at Trimble, said, “4D imaging radar could be a game changer for jobsite solutions. We are continuing to evaluate the capabilities for high resolution sensing in these harsh environments.” 

DPR’s Roedel sums up the construction industry’s technology directions: “Combining AI with automation could be the 2024 advancement that will truly revolutionize our industry. It’s likely that this combination will challenge the traditional hierarchical management structures that have been in place for decades, setting the stage for truly streamlined workflows.”