They create 'cannibal' robots that change size and grow stronger by consuming other machines.

Robots are machines: they don't grow or get stronger unless someone decides to add or improve their parts. They don't eat like humans, nor do they evolve over time like we do. Or at least, that's how it's been until now. Scientists at Columbia University (USA) have developed a technology that allows robots to do something that until now seemed exclusive to living organisms: grow using materials from their environment. They call this "robotic metabolism," and it could be the key to enabling machines to mutate, grow, and improve on their own, without the need for human intervention.

"True autonomy means that robots must not only think for themselves, but also be physically self-sufficient," explains Philippe Martin Wyder, senior author of the study recently published in the journal Science Advances , and a professor of engineering at Columbia University. "Just as biological life absorbs and integrates resources, these robots grow, adapt, and repair themselves using materials from their environment or from other robots," he adds.

The system is based on components called Truss Links : robotic bars that can be lengthened or shortened and connected to each other by magnets. These components are assembled into shapes such as triangles, stars, and three-dimensional structures called tetrahedrons. What's new is that, through centralized control, these shapes can be combined, divided, or transformed into more complex ones. For example, a star-shaped shape can be folded into a triangle and transformed into a more stable tetrahedron.

Indeed, in the experiments, the researchers demonstrated that a structure could incorporate additional parts to increase its size or improve its mobility. They also showed that a shape could be divided and then rearranged to return to its original configuration. All of this was done in a controlled environment, with movements choreographed by a central system, not completely autonomously.

That is, the robots don't make decisions on their own or actively search for parts. For now, the entire process requires researchers to place the modules and direct their actions from a computer. However, the potential is enormous. In the future, with sensors, decentralized control, and machine learning, these machines could become much more independent.

What are the implications of this breakthrough? If perfected, according to the researchers, it could allow for the use of mutating robots that could be especially useful in situations where human intervention is difficult or dangerous: for example, in areas affected by natural disasters, on space missions, or in the exploration of extreme environments.

Although the system is still in the experimental phase and its capabilities are limited, it marks an important step toward a new generation of robots with bodies capable of changing. Truss Links are not strictly living machines, but they do exhibit physical behavior more similar to that of living beings thanks to their ability to grow and transform through the use of other parts, all while minimizing human intervention. In the future, the sky's the limit.

"The image of self-replicating robots conjures up some unsavory science fiction scenarios," says Hod Lipson, co-author of the study and head of the Department of Mechanical Engineering at Columbia University. "But the reality is that as we delegate more tasks to robots—from self-driving cars to automated manufacturing, and even defense and space exploration—who will take care of these robots? We can't rely on humans to maintain these machines. Robots must ultimately learn to take care of themselves," the researcher concludes.