Why Construction’s Robot Revolution Will Creep, Not Crash

The robot revolution won’t march on to a construction site in a single, hard-hatted invasion. It will creep in. One task will shift. An imperfect machine will take over part of a workflow, perhaps with a human ready to intervene. Then another will follow. This is the grounded half of the message from the Queensland University of Technology (QUT) Centre for Robotics director Professor Michael Milford from his opening speech at The Urban Developer’s flagship conference Urbanity-26 on the Gold Coast. The other half is considerably more dramatic. “Almost anything is now distinctly possible in robotics and automation,” Milford says. “No responsible roboticist or technologist would have made this statement four years ago.” Possible doesn’t mean viable Milford attributes the shift to two converging forces: modern artificial intelligence and rapidly advancing, increasingly affordable hardware. Artificial intelligence allows robots to behave as though they possess common sense and interact more naturally with people. At QUT, a robot instructed simply to place something healthy on a colleague’s desk could identify fruit as a suitable snack, infer where to find it and navigate towards the office kitchen. “Robots could not do this five years ago,” Milford says. Meanwhile, some robotic hardware now costs about one-hundredth of its equivalent a decade or more ago. ▲ QUT Centre for Robotics director Professor Michael Milford: Almost anything is now distinctly possible in robotics and automation. Milford says more money flowed into robotics start-ups during the first half of 2026 than in any previous full year. But possibility isn’t viability. Nor is an impressive demonstration proof of a commercially useful product. Robots able to perform flips or martial arts may still struggle to pick up ordinary objects. A machine can excel spectacularly in one area and fail spectacularly in another. Progress, Milford warns, would be “grossly uneven”. Commercial adoption instead rests on a less cinematic calculation: how well must the technology perform, what happens when it fails and can a human step in? A delivery robot can stop and summon a remote operator. An autonomous vehicle facing a life-or-death decision in a fraction of a second cannot. A household cleaning robot may still offer value despite making mistakes. A passenger-carrying robotaxi must operate almost perfectly. “These things always take longer than people think,” Milford says, pointing to autonomous vehicles’ 40-year journey from early demonstrations to emerging commercial deployment. Construction’s quiet takeover Development and construction offer an enormous prize for robotics companies, but the experience of other large industries shows scale alone does not guarantee adoption. Automation has permeated mining while making few comparable inroads into agriculture, despite its considerable potential. Aged care is only beginning to move. For property, one leading indicator of faster change may be a swarm of what Milford jokingly described as the “most obnoxious 18 and 19-year-old startup founders beating down your door constantly” with new products. Such attention would signal a surge of capital and technical talent targeting the sector. Even then, Milford expects automation to enter property in pieces. Other industries have not torn apart entire operating models and started again. Robots and software have slipped into established processes, taking over an individual function or refining one step before moving to the next. Construction’s robot revolution, in other words, is more likely to creep through the gate than crash through it. Don’t rebuild the city for the machine Milford offers developers and city makers a simple test for judging urban technology. “If you have to substantially change the physical infrastructure, the layout of the city in order for the technology to work, then it’s probably a technology you shouldn’t be considering adopting.” ▲ Milford says technology that needs substantial infrastructure changes, such as dedicated lanes for autonomous vehicles, is not a technology worth adopting. Proposals for dedicated autonomous-vehicle lanes illustrate the problem. The companies pursuing driverless transport generally want systems requiring as little new infrastructure, government support and regulatory intervention as possible. The same principle applies inside property businesses. Useful technology should assimilate into existing operations and make them more economically viable. A product requiring a developer to reorganise an entire company around it begins with an immediate commercial disadvantage. Barriers to experimentation, however, have fallen sharply. Even smaller operators can now trial AI tools cheaply inside controlled environments without committing large teams or budgets. Milford’s advice isn ’ t to implement everything. Early adopters have already invested tens, perhaps hundreds, of billions of dollars globally without receiving the efficiency gains they expected, he says. Nor should disappointment with immature products be mistaken for evidence of limited long-term potential. Instead, property leaders should “keep your hands sufficiently dirty”, building enough practical knowledge to understand what AI can do now, where it fails and what may become commercially useful within several years. Such experience provides the best defence when a vendor arrives with a shiny product and an expensive promise. The humanoid wildcard Humanoid robots could upset Milford’s incremental trajectory. The physical world already accommodates bodies with two arms, two legs and human proportions. A capable humanoid machine could therefore enter existing buildings and workplaces without forcing them to be redesigned. If the technology and its human factors advance far enough, Milford raises the possibility of millions of humanoid robots appearing rapidly, followed by billions within five or 10 years. He is careful to distinguish possibility from prediction. “I’m not saying it is going to happen, but it’s now a tangible possibility,” he says. Technology will not solve Australia’s housing pressures by itself, or provide most of the answer. Milford nevertheless sees tantalising opportunities to improve affordability, expand development capacity and support mobility. For property leaders, the immediate question isn ’t when robots take over. It ’ s which small, commercially viable task changes first—and who recognises its value before everyone else does.

Exclusive
Vanessa Croll
Thu 23 Jul 26

Why Construction’s Robot Revolution Will Creep, Not Crash

Get the full story.