If one robot breaks down, the others keep working.
In a laboratory at Adelaide University, researchers have drawn on the ancient intelligence of bees and ants to imagine a new kind of mining — one where small, autonomous robots think collectively, adapt without instruction, and carry on even when one of their own falls. The work, led by Dr Noune Melkoumian and doctoral researcher Dr Joven Tan, points toward a future where the most dangerous places on Earth — and perhaps beyond it — need not be entered by human hands at all. It is a reminder that some of nature's oldest solutions may yet answer some of industry's newest problems.
- Mining is pushing into deeper, more dangerous territory, and existing automated systems are too rigid and too dependent on centralised control to keep pace.
- A honeybee-inspired swarm of hand-sized robots cut travel distance by 80%, halved energy use, and finished tasks 60% faster than conventional approaches in laboratory trials.
- The swarm's defining strength is resilience — if one robot fails, the others continue, eliminating the single points of failure that cripple traditional automated systems.
- Sensors, battery life, and the raw unpredictability of real underground environments remain unsolved challenges standing between the laboratory and a working mine.
- The technology's horizon stretches beyond Earth — autonomous swarms capable of collective decision-making are seen as a natural fit for asteroid and lunar mining where human oversight is impossible.
In a laboratory at Adelaide University, small robots move across a simulated mining floor without waiting for orders. They work the way bees work — each making its own decisions, each contributing to a shared goal, each capable of continuing if others fail. It is the foundation of a swarm robotics system that researchers believe could change how the world extracts minerals from the earth.
The team, led by Dr Noune Melkoumian and doctoral researcher Dr Joven Tan, spent months studying how social insects solve problems collectively. Ants divide labour between scouts and carriers. Honeybees map territory before they harvest. The researchers asked what mining might look like if it followed the same logic — dozens of small machines communicating and adapting, rather than a single computer issuing commands.
They tested three systems using Zumo 2040 robots in a controlled environment mimicking underground conditions. A basic system had robots collect ore immediately. An ant-inspired model split scouting from transport. A honeybee model sent robots to explore and map before any collection began. The honeybee approach proved most powerful: 80% less travel, 50% less energy, tasks completed 60% faster. More importantly, the swarm was resilient — no single breakdown could stop the whole operation.
The stakes are practical. Mining is moving deeper underground and into more remote, hazardous locations. Existing automation helps, but it is expensive, inflexible, and dependent on infrastructure that real mines often cannot support. Swarm robots could work in spaces too tight or toxic for people, making decisions without waiting for wireless signals or central servers.
Dr Melkoumian was careful to note that the team built and observed physical robots rather than relying on simulation alone — closing some of the gap between theory and practice. But she was equally clear about what remains: tougher sensors, longer battery life, and the ability to handle the collapses, water, and electromagnetic chaos of a real working mine.
The path forward, the researchers believe, leads first to the most dangerous mining zones on Earth, and eventually to space — where asteroids and the moon hold vast resources that only fully autonomous systems could ever reach. For now, the work continues in the laboratory, where small robots learn from millions of years of insect evolution.
In a laboratory at Adelaide University, small robots no bigger than a hand move across a simulated mining floor without waiting for orders from a central command. They work together the way bees work together—each one making its own decisions, each one contributing to a shared goal, each one capable of continuing the job even if others fail. This is the premise behind a new swarm robotics system that researchers believe could reshape how we extract minerals from the earth.
The team, led by Dr Noune Melkoumian at Adelaide's School of Chemical Engineering, spent months studying how social insects solve problems collectively. Ants divide labor: some scout for food while others carry it home. Honeybees map territory before they harvest. The researchers asked a simple question: what if robots could work the same way? What if mining operations didn't need a single computer making all the decisions, but instead relied on dozens of small machines that communicated with each other and adapted on the fly?
To test the idea, they built three different systems using Zumo 2040 robots in a controlled environment designed to mimic underground mining conditions. The first was basic—robots found ore and brought it back immediately. The second borrowed from ant behavior, splitting the work: one robot scouted while another transported. The third, inspired by honeybees, sent robots out first to explore and map the space before any collection began. The difference in results was striking. The honeybee approach cut travel distance by 80 percent, slashed energy use by half, and completed tasks 60 percent faster than the baseline system. The ant-inspired model also showed gains by letting robots specialize.
Dr Joven Tan, who led the research as part of his doctoral work, saw in these results something deeper than efficiency metrics. "Social insects have developed very efficient ways of solving problems together," he said. "By applying these ideas to robotics, we can create systems that are more efficient, adaptable and reliable for industries such as mining." The key advantage wasn't just speed or power savings. It was resilience. If one robot broke down, the others kept working. If conditions changed, the swarm adapted. No single point of failure could shut down the entire operation.
This matters because mining is moving into harder places. Operations are going deeper underground, into more remote locations, into environments where human workers face constant danger. Automation has already improved safety and productivity, but existing systems are expensive, rigid, and dependent on infrastructure that doesn't always survive the chaos of a real mine. A swarm of autonomous robots could work in spaces too tight or too toxic for people, making decisions without waiting for wireless signals or central servers.
Dr Melkoumian emphasized that the research team didn't rely on computer simulations alone. They built physical robots and watched them work in a real laboratory setting. That matters because the gap between theory and practice in robotics is often wide. But she was also clear about what still needs to happen. The sensors need to be tougher. The batteries need to last longer. The systems need to handle the unpredictability of actual underground environments—the collapses, the water, the electromagnetic interference, the sheer chaos of a working mine.
Even so, the researchers see a clear path forward. Swarm robotics could eventually operate in the most hazardous mining zones, pulling workers out of danger while maintaining or improving productivity. And the technology might go further still. Space mining—extracting resources from asteroids or the moon—will require autonomous systems that can work without constant human oversight. A swarm of robots that can think together, adapt together, and survive together might be exactly what that future demands. For now, though, the work continues in the laboratory, where small robots move across a simulated floor, learning from millions of years of insect evolution.
Notable Quotes
Social insects have developed very efficient ways of solving problems together. By applying these ideas to robotics, we can create systems that are more efficient, adaptable and reliable for industries such as mining.— Dr Joven Tan, Adelaide University School of Chemical Engineering
Nature has spent millions of years developing efficient ways for groups to work together. By learning from these systems, we can develop new technologies that are more flexible, reliable and efficient.— Dr Noune Melkoumian, project leader
The Hearth Conversation Another angle on the story
Why does it matter that these robots don't have a central controller? Couldn't you just make the central system smarter?
You could, but then you've created a single point of failure. If that central computer breaks or loses signal—which happens in mines—everything stops. A swarm keeps working because each robot carries its own intelligence. It's redundancy built into the design itself.
The honeybee approach was fastest. But why would bees have evolved that way? What problem were they solving?
Bees need to minimize wasted trips. If a scout finds flowers far away, it makes sense to map the whole area first, then send the whole colony out efficiently. In mining, that same logic saves energy and time. You explore once, then you harvest.
You mentioned these robots are small. How small are we talking?
Zumo 2040s—about the size of your hand. In a real mine, you'd probably deploy dozens or hundreds of them. The idea is that you don't need one massive machine. You need many small ones that can squeeze into tight spaces and work around obstacles.
What happens when a robot gets stuck or damaged underground?
The swarm keeps going. That's the whole point. In a traditional system, you'd have to stop everything, send someone down to fix it, restart the operation. With a swarm, you lose one unit and the others compensate. The job continues.
How far away is this from actually being used in a real mine?
Years, probably. The laboratory is controlled. A real mine is wet, unstable, electromagnetically noisy. The sensors need to be more robust, the batteries need to last longer, and the robots need to handle terrain that's nothing like a flat lab floor. But the proof of concept is solid.
Could this technology work in other industries?
Almost certainly. Any job that's dangerous, repetitive, and involves teams of workers could benefit—disaster response, deep-sea salvage, even manufacturing. But mining is the obvious first application because the conditions are so harsh and the safety stakes are so high.