Humanoid robots are learning to dance. But can we trust them at work?

Professor Jonathan Roberts, QUT Centre for Robotics and QUT School of Electrical Engineering & Robotics, and Professor Glenda Caldwell, QUT School of Architecture & Built Environment

Humanoid robots can now walk, climb stairs, carry objects and even perform coordinated dance routines. Videos of robots doing backflips and martial arts have attracted millions of views online.

The progress is remarkable. But watching a robot perform a backflip is one thing. Having one work alongside you in a hospital, on a construction site or in a factory is quite another.

The important question is no longer simply what humanoid robots can do. It is what we should allow them to do, and under what conditions.

Why humanoids, and why now?

Australia faces some difficult challenges. Our population is ageing, industries struggle to find workers, and productivity growth has been disappointing.

Humanoid robots could help address some of these problems.

Unlike traditional industrial robots, which generally operate in structured environments, humanoids are designed to move through spaces built for people.

They could assist care workers with physically demanding tasks, move materials around construction sites or support manufacturing operations where conventional automation is difficult to implement.

Rather than redesigning workplaces around machines, we might eventually introduce machines that can operate within existing environments.

But there is an enormous difference between demonstrating a capability and performing it reliably, safely and economically in a real workplace.

Technology moves faster than regulation

History shows that new technologies can enter society before appropriate rules and safeguards are established. This is sometimes called the pacing problem.

Industrial robots, ridesharing platforms and, more recently, artificial intelligence have all challenged existing regulatory frameworks.

Humanoid robots could present an even greater challenge because they bring artificial intelligence into the physical world.

They will not just generate information. They will move, manipulate objects and potentially act directly on people and their surroundings.

We therefore need to consider questions about safety, responsibility, privacy and workplace rights before humanoids become commonplace.

Waiting until problems emerge risks repeating mistakes made with earlier technologies.

Australia's opportunity

China, the United States and other countries are investing heavily in humanoid development.

Australia is unlikely to compete in mass-producing humanoid hardware. But we have an opportunity to lead in something equally important: understanding how humanoids can be introduced safely, ethically and effectively into workplaces.

This is the focus of the new Australian Humanoid Hub, bringing together researchers from QUT, Swinburne University of Technology and Flinders University, alongside industry and government partners.

Building on Australia's established expertise in robotics, the Hub will investigate humanoid adoption in care, construction, manufacturing and logistics.

Its work will range from testing robot capabilities in realistic environments to examining workplace safety, regulation, job design and human–robot interaction.

The aim is not to demonstrate that humanoids can do everything people can do.

It is to establish where they genuinely provide benefits, where the risks outweigh those benefits, and how decisions about their use should be made.

People must remain at the centre

Humanoid robots offer exciting possibilities, particularly in industries facing labour shortages and growing demand.

But technological capability alone is not a sufficient reason to introduce a robot into a workplace.

We need evidence that humanoids improve safety, productivity and people's lives.

That means experimenting, asking difficult questions, involving workers and communities, and being prepared to conclude that some applications are simply not appropriate.

The future of humanoid robotics should not be determined solely by what engineers can build or manufacturers can sell.

It should be shaped by the people who will live and work alongside these machines.

About the researchers

Professor Jonathan Roberts, from the QUT Centre of Robotics, is the director of the ARC Research Hub for Safe and Ethical Humanoid Robot Implementation and Adoption.  His main area of research is in the area of how to enable people to work with robots. Professor Glenda Caldwell, deputy director of the hub, her research area is on how we design spaces and processes for people to interact and collaborate with robots.