Robot Shares Power Between Modules, Beats System Failures

Scientists just solved one of robotics' biggest headaches by taking a page from nature's playbook.

Researchers at Switzerland's EPFL have built a robot that actually becomes more reliable when it has more parts. That flips conventional wisdom on its head, where adding components usually means more things can break down.

The breakthrough centers on something the team calls hyper-redundancy. The Mori3 robot's four triangular modules share everything: battery power, wireless communication, and sensor data. When one module completely dies, its neighbors keep it running by lending their resources.

Lead researcher Jamie Paik put it simply: "For the first time, we have found a way to reverse the trend of increasing odds of failure with increasing function." Her team drew inspiration from flocking birds that share sensing information and trees that warn neighbors of threats through airborne signals.

The real test came when researchers deliberately killed the robot's central module by cutting its power, communication, and sensors. Normally, that dead module would paralyze the entire system. Instead, the Mori3 successfully walked toward a barrier and squeezed underneath it, with working modules compensating completely for their failed teammate.

Previous modular robots tried backup resources or self-reconfiguration when parts failed, but those fixes rarely restored full functionality. The EPFL team discovered that sharing just one or two resources wasn't enough either. Only when all three critical resources flowed freely between modules did the system become more resilient with size.

The Ripple Effect

This resource-sharing framework could transform how we design everything from search and rescue robots to space exploration systems. Robots operating in dangerous or remote locations often can't be repaired quickly, making reliability crucial.

The methodology essentially revives dead modules within a collective, bringing them back to full functionality. That resolves what roboticists call the reliability-adaptability conflict, where making robots more capable traditionally made them more fragile.

The team plans to scale up their approach to robotic swarms, where dozens or hundreds of units could dock together to transfer energy and information. Picture disaster response robots that keep working despite damaged units, or planetary rovers that support each other across hostile terrain.

First author Kevin Holdcroft sees the potential clearly: "Our local resource-sharing framework has the potential to support highly adaptive robots that can operate with unprecedented reliability." The research appears in Science Robotics, marking a significant step toward robots that work more like living organisms and less like breakable machines.