Modular Robots Boost Resilience by Sharing Resources
- tech360.tv
- 2 minutes ago
- 3 min read
Roboticists at EPFL have demonstrated that modular robots can become significantly more resistant to failure by sharing power, sensing, and communication resources among individual units. This approach reverses the trend where traditional robotic systems often lose functionality if one element breaks down.
Minimising the possibility of failure is a top priority in robotic design, particularly as systems with multiple units can perform more diverse functions. However, more units historically meant more parts that could potentially fail.
Researchers, led by Jamie Paik, head of the Reconfigurable Robotics Laboratory (RRL) at EPFL’s School of Engineering, have designed a modular robot that actually lowers its odds of failure through resource sharing. Paik explained that this marks the first time a method has been found to reverse the trend of increasing failure odds with more functionality.
“We introduce local resource sharing as a new paradigm in robotics, reducing the failure rate with a larger number of modules,” Paik stated. The team published their findings in Science Robotics.
Their paper showed how a modular origami robot successfully navigated complex terrain, even with one module completely deprived of power, sensing, and wireless communication. This was possible by exploiting redundant resources and sharing them locally.
The RRL team drew inspiration from nature, where collective solutions often address failure. Birds share local sensing information through flocking behaviour, while some trees communicate threats to neighbours using airborne signals. Cells continuously transport nutrients across their membranes so that the death of any individual doesn't significantly impact the overall organism.
Modular robots, composed of multiple connecting units, are analogous to multicellular or collective organisms. Until now, their design presented a vulnerability, as a single module's failure could disable some or all of the robot's tasks.
Some modular robots feature built-in backup resources or self-reconfiguration abilities, but these often do not completely restore functionality. For their study, the RRL team utilised hyper-redundancy, involving the sharing of all critical power, communication, and sensing resources across all modules without altering the robot’s physical structure.
Paik noted that sharing only one or two resources was insufficient. If each resource had an equal chance of failure, system reliability would continue to drop with more agents. However, when all resources were shared, this trend was reversed.
In a locomotion experiment using the Mori3 robot, which consists of four triangular modules, the team cut battery power, wireless communication, and sensing to the central module. Normally, this "dead" central module would impede the articulation and movement of the other three.
Thanks to hyper-redundancy, the neighbouring modules fully compensated for the central module’s lack of resources. This enabled the Mori3 to "walk" toward a barrier and contort itself effectively to pass underneath.
RRL researcher Kevin Holdcroft, the first author, summarised, “Essentially, our methodology allowed us to 'revive' a dead module in a collective and bring it back to full functionality.” He added that the local resource-sharing framework has the potential to support highly adaptive robots that operate with unprecedented reliability, resolving the reliability-adaptability conflict.
Future work could focus on applying this resource-sharing framework to more complex systems with an increasing number of agents. The concept may extend to robotic swarms, with hardware adaptations allowing swarm members to dock for energy and information transfer.
Modular robots can increase resilience by sharing power, sensing, and communication resources.
This resource-sharing approach reverses the trend of increased failure with more robot modules.
The Mori3 robot successfully performed tasks even when a central module was deprived of all resources.
Source: TECHXPLORE