Moving Beyond Prototypes
Autonomous heavy machinery is finally doing actual work on active jobsites, rather than just running demonstrations in controlled sandboxes. Bedrock Robotics, working alongside Sundt Construction and Zachry Construction, has put its autonomous excavators to work on a 130-acre manufacturing facility in the Southwest. The site is being prepped to support domestic energy production, and the numbers are substantial. The autonomous units have already moved over 65,000 cubic yards of earth and rock. This isn't a pilot program; it's a commercial deployment proving that self-driving construction tech can handle the unpredictable nature of heavy civil site prep.
What makes this interesting is the approach Bedrock has taken. Founded by engineers from Waymo, the company didn't try to reinvent the excavator. Instead, they built a retrofit system. They take existing machinery, ranging from 20-ton to 80-ton models, and install their proprietary array of LiDAR, GPS, and computing hardware. This strategy drastically lowers the barrier to entry for contractors. You don't have to buy a fleet of futuristic robots; you just upgrade the iron you already own. It's a pragmatic approach to a complex engineering problem, focusing on immediate utility rather than distant, sci-fi visions of construction.
The Hybrid Workflow Advantage
The most critical aspect of this deployment is how the autonomous machines interact with human crews. The Bedrock excavators don't operate in an isolated zone. They load human-driven articulated dump trucks as part of a standard load-and-haul cycle. A truck pulls up; the autonomous excavator senses its position, calculates the optimal swing path, scoops from a stripped pile, and loads the bed.
We observed several practical reasons why this hybrid model is gaining traction:
•Zero Disruption to Site Logistics: The autonomous units slot directly into existing traffic patterns. Site managers don't have to redesign their entire operation to accommodate the technology.
•Scalable Deployment: Contractors can start with a single retrofitted machine, test its integration with their human drivers, and scale up only when they see a clear return on investment.
•Immediate Productivity: Because the fundamental workflow remains identical to manual operations, the site sees immediate benefits from the continuous, fatigue-free operation of the excavator.
Construction sites are chaotic, highly coordinated environments. Introducing a completely new operational paradigm usually results in massive friction and lost productivity. By keeping the load-and-haul cycle exactly as it has always been, Bedrock allows contractors to adopt autonomy incrementally. The 65,000-yard milestone proves that this hybrid approach isn't just theoretical; it maintains the required pace of a major industrial project.
The Operator Deficit Reality
The push for autonomy isn't just about efficiency; it's an existential necessity for the industry. Finding skilled operators for heavy machinery, particularly for remote or massive civil projects, is becoming increasingly difficult. The workforce is ageing out, and new entrants aren't replacing them fast enough.
Boris Sofman, CEO of Bedrock Robotics, pinpointed this exact issue. "Just the availability of labor is a huge pain point, so a lot of jobs just don't get done," Sofman noted, describing the technology as having an "expansionary impact on construction."
When a contractor can't find enough operators, projects get delayed, bids are rejected, and growth stalls. The industry needs nearly half a million new workers just to meet current demand. Autonomous equipment offers a way to bypass this bottleneck. It allows firms to bid on larger projects without the prerequisite of expanding their human workforce. The Sundt deployment illustrates this perfectly. By automating a portion of the earthmoving, the human operators can be reassigned to more complex, nuanced tasks that require human judgement, effectively multiplying the output of the existing crew.
Training AI with Human Expertise
One of the most fascinating elements of Bedrock's system is how it learns. It doesn't rely solely on abstract algorithms; it learns from the best human operators on the site. The process involves recording the inputs and decisions of experienced professionals as they handle different soil types, rock formations, and loading scenarios.
Dan Green, Sundt Senior Project Manager, explained the value of this method. "We've taken very experienced operators and they're teaching how a human does that task by using multiple operators," Green said. "It's getting to feel the difference between them, because every operator is different in how they do their job."
By aggregating these diverse techniques, the system develops a robust operational model. As Green pointed out, "Using these autonomous systems efficiency, you're kind of taking that consolidated knowledge of multiple operators and turning it into one operator, and that's very powerful."
This approach solves a major problem in robotics: handling edge cases. An operator with twenty years of experience knows instinctively how to adjust their swing when hitting a hidden rock shelf or how to feather the controls in loose sand. By feeding this specific, hard-won expertise into the machine learning model, Bedrock creates an autonomous system that doesn't just execute a pre-programmed path, but reacts dynamically to changing site conditions based on consolidated human wisdom.
Scaling Up on Active Sites
The 65,000 yards moved so far is just the beginning. The total scope of the manufacturing facility project involves moving approximately 700,000 cubic yards of material.
"They're a tool in a much larger toolbox that we have on this," Green stated regarding the autonomous units. "That's the scale of what we're moving. Bedrock excavators are about 10 percent of our utilization out there."
Bedrock is actively expanding its footprint, partnering with firms like Austin Bridge & Road, Maverick Constructors, and Haydon Cos. to test its systems across diverse environments. This strategy of developing the technology on active jobsites, rather than in sterile test facilities, is crucial. It forces the engineering team to solve real-world problems like unpredictable topography, dynamic obstacle avoidance, and integration with varied fleet compositions.
"We've been really happy with how much a pretty sizable group, a part of the industry is embracing the possibility of this," Sofman added. The willingness of major contractors to put these systems on critical path projects indicates a significant shift in industry confidence.
Looking Ahead
The deployment in Texas marks a transition point. Autonomous earthmoving is no longer a science project; it's a deployable asset. While it currently represents a fraction of the total utilisation on this specific site, it proves the concept works at scale. The future of heavy civil construction will likely be defined by this hybrid approach, where autonomous machines handle the repetitive, high-volume tasks, allowing skilled human operators to focus on precision work and site management.
The Economic Imperative of Automation: The push towards autonomous earthmoving is fundamentally driven by economics. The traditional model of heavy civil construction relies heavily on a predictable supply of skilled labour at predictable rates. However, the current market dynamics have shattered this predictability. When a contractor bids on a multi-year infrastructure project, they must estimate labour costs years in advance. If a local labour shortage forces them to bring in operators from out of state, paying premium wages and per diems, the project's profit margins evaporate rapidly.
Autonomous systems like those deployed by Bedrock provide a hedge against this wage volatility. While the initial capital expenditure for the retrofit kit is significant, the ongoing operational cost becomes highly predictable. An autonomous excavator does not require overtime pay, it does not need breaks, and its "wage" is essentially the cost of fuel, maintenance, and software licensing. This predictability allows contractors to bid more aggressively and manage risk more effectively on large-scale, long-duration projects.
Furthermore, the ability to operate equipment continuously, potentially 24 hours a day in remote locations, drastically compresses the project schedule. In heavy civil construction, time is directly correlated with money; finishing site prep weeks ahead of schedule reduces overhead costs and accelerates the timeline for subsequent trades, creating a cascading economic benefit across the entire project lifecycle.
Takeaway
•Commercial Viability Proven: Moving 65,000 cubic yards of material on an active 130-acre site demonstrates that autonomous excavators are ready for heavy civil deployment.
•Retrofit Strategy Lowers Barriers: By upgrading existing 20- to 80-ton excavators rather than building new machines, Bedrock allows contractors to leverage their current fleet investments.
•Hybrid Workflows Work: Integrating autonomous loaders with human-driven dump trucks provides immediate productivity gains without requiring a redesign of site logistics.
•Consolidated Expertise: Training the AI using the combined techniques of multiple experienced operators creates a system capable of handling complex, variable site conditions.
•Mitigating the Labour Shortage: Autonomy provides an "expansionary impact," allowing contractors to take on larger projects despite the severe shortage of skilled heavy equipment operators.
Call to Action
To learn more about how autonomous technologies and artificial intelligence are reshaping project delivery and construction logistics, subscribe to the Project Flux newsletter.
Links and Stuff
All content reflects our personal views and is not intended as professional advice or to represent any organisation.
/
1