Imagine a robot that keeps working even when part of it fails—a robot that can literally share its energy, senses, and information with its broken parts to continue performing its tasks. Thanks to researchers at the École Polytechnique Fédérale de Lausanne (EPFL), this vision is now a reality. Their breakthrough in modular robotics promises machines that are far more resilient than anything built before.
Traditional robots are vulnerable. If a single component fails, the robot can stop working or perform poorly. This limitation is especially significant for robots made up of multiple units or “modules.” While multiple modules allow a robot to perform a wider variety of functions, they also create more points of failure. A broken part can disrupt the entire system. This has been one of the biggest challenges in robotic design: balancing functionality with reliability.
But EPFL’s Reconfigurable Robotics Laboratory (RRL), led by Jamie Paik, has developed a solution that flips this problem on its head. Instead of avoiding failure-prone designs, the team designed a robot that becomes more reliable as it gains more modules. Their key innovation? Local resource sharing.
Sharing Is the Secret
In nature, many organisms overcome the problem of individual failure by working together. Birds in a flock share sensory information to avoid predators. Trees send warning signals to neighbors when under threat. Cells in our body constantly move nutrients so that the death of one cell does not compromise the organism.
Inspired by these natural systems, EPFL’s roboticists created a modular robot that can share all critical resources—power, sensing, and communication—among its modules. This approach, called hyper-redundancy, ensures that if one module loses its resources, the other modules can compensate, keeping the entire robot functional.
“For the first time, we have found a way to reverse the trend of increasing odds of failure with increasing function,” says Paik. “We introduce local resource sharing as a new paradigm in robotics, reducing the failure rate with a larger number of modules.”
Testing the Concept
To demonstrate their concept, the team used a modular robot called Mori3, composed of four triangular units. Normally, if the central module lost its power, communication, or sensing ability, the robot would be unable to move effectively. But when the RRL team applied hyper-redundancy, the neighboring modules fully compensated for the missing resources. The “dead” central module was effectively revived, and the Mori3 successfully navigated a complex obstacle course. It could walk toward a barrier and contort itself to pass underneath, something impossible with a non-resource-sharing design.
Kevin Holdcroft, first author of the study, explains: “Essentially, our methodology allowed us to ‘revive’ a dead module in a collective and bring it back to full functionality. Our local resource-sharing framework therefore has the potential to support highly adaptive robots that can operate with unprecedented reliability, finally resolving the reliability-adaptability conflict.”
The researchers also discovered that partial sharing was not enough. Sharing only one or two resources—like power or communication alone—did not significantly improve reliability. To fully reverse the failure trend, all critical resources needed to be shared across the robot.
Why It Matters
This breakthrough could transform how robots are built and deployed. Currently, most modular robots have built-in backup systems or self-reconfiguration abilities to cope with failure. While helpful, these approaches rarely restore full functionality. EPFL’s hyper-redundancy framework goes further by ensuring that a failed module can be completely compensated for, without altering the robot’s physical design.
This design has several potential applications:
Exploration in Hazardous Environments: Modular robots could operate reliably in dangerous conditions where failures are likely, such as deep-sea exploration, disaster zones, or extraterrestrial landscapes.
Medical Robots: Swarms of tiny surgical or diagnostic robots could continue functioning even if some units fail, increasing patient safety.
Industrial Automation: Factories using modular robots could reduce downtime caused by individual component failures.
Looking Ahead: From Robots to Swarms
EPFL researchers are now considering how to extend this resource-sharing concept to larger and more complex systems. One exciting possibility is robotic swarms, where multiple independent robots can dock together to exchange power and information. By doing so, swarms could maintain functionality even when some robots are incapacitated, making them incredibly resilient collective systems.
Paik highlights that this approach does not require changes to the robot’s physical structure, which is a major advantage. The hyper-redundant design is purely based on how the modules manage and share their resources, meaning it can be applied to existing robotic platforms with minimal modifications.
The Future of Resilient Robotics
This work, published in Science Robotics under the title “Scalable robot collective resilience by sharing resources”, represents a major step forward in robotic engineering. By drawing inspiration from natural systems and applying it to modular robots, EPFL researchers have demonstrated a way to create machines that are both highly functional and remarkably resilient.
As robotic systems become more complex and are tasked with critical jobs in unpredictable environments, such innovations will be crucial. Hyper-redundancy could allow robots to work alongside humans more safely, explore extreme environments more effectively, and even collaborate in ways we have not yet imagined.
In essence, the EPFL team has shown that robots can be more than just machines—they can be adaptive collectives, learning from nature’s strategies to overcome failure and maximize performance. This could mark the beginning of a new era where robots are no longer fragile systems prone to breakdown, but resilient collaborators capable of thriving even in challenging situations.
Reference: Kevin Holdcroft et al, Scalable robot collective resilience by sharing resources, Science Robotics (2026). DOI: 10.1126/scirobotics.ady6304

Comments
Post a Comment