By Syl Kacapyr
In a groundbreaking exploration of robotics, Cornell engineers have unveiled a remarkable innovation—the Cross-Link Collective. This system of small, modular robots operates in a manner reminiscent of flowing materials rather than traditional mechanical devices. The essence of this innovation lies in its ability to adapt, reshape, and respond to its environment without a centralized control system, marking a significant leap in robotic technology.
The study, published in *Science Robotics*, showcases how this collective of dozens of small robots with limited individual mobility can weave together to perform coordinated tasks, demonstrating a form of behavior akin to soft matter. This concept of “mechanical intelligence” allows the robotic system to deform and reorganize dynamically, ultimately achieving seamless motion through collaboration.
According to Kirstin Petersen, the project’s leading researcher and a professor at Cornell, the focus has shifted from relying solely on computation and communication to leveraging the physical interactions between robots. “We’re using the very shape of the robots and their interactions to drive behaviors,” Petersen explains. This innovation allows the collective to settle into configurations that minimize internal stresses while enhancing their motion capabilities.
Each robotic module in the Cross-Link Collective is compact, measuring approximately 200 millimeters long and 20 millimeters wide. Inside each module is a small motor that facilitates oscillation between two distinct shapes, “I” and “U.” This oscillation enables the robots to exert forces against the ground, inching forward and making contact with neighboring modules. With Velcro patches at either end, these robots can temporarily latch onto one another, allowing for flexible and dynamic movement.
On an individual level, these modules may seem sluggish and inefficient. However, when they form chains and collaborate, they harness their collective strength, self-organizing in ways that allow them to navigate complex environments with resilience. For instance, when traversing inclined surfaces, these interconnected chains proved significantly more efficient than isolated robots, which often found themselves stalled based on their orientation.
In more challenging terrains, the collective functions like a fluid, maintaining cohesion through the formation and dissolution of connections to avoid jamming. This adaptive quality ensures that the system remains functional even if a single module encounters a failure, like a depleted battery. “The system is inherently redundant, allowing it to adapt and function without relying on any one module,” notes Danna Ma, a visiting lecturer and co-author of the study.
Interestingly, the researchers found that minimal computation could enhance the system’s performance. For example, if an isolated robot starts to fall behind, it can emit an audible distress signal. This prompts neighboring modules to adjust their speed, allowing the slower robot to reconnect. “There’s no centralized control,” explains Ma. “Each module can detect when it’s out of sync simply by the degree of jostling it experiences, facilitating a simple yet effective communication method.”
The original design of the robotic modules was developed by co-authors at the Georgia Institute of Technology, while Petersen and Ma have spent years refining the system through experimentation and statistical analysis. This comprehensive process highlighted how even minor modifications in the modules’ size and characteristics could significantly affect how effectively they join together and operate as a cohesive unit.
Inspired by active gels—materials that continuously form and dissolve their molecular links—the Cross-Link Collective not only provides insights into robotics but also offers potential applications in soft-matter engineering. Petersen emphasizes the importance of considering what can be encoded into the physics of a system. “As robots become more integrated into real-world applications that present dynamic challenges, relinquishing tight control can surprisingly yield useful emergent behaviors,” she notes.
Cornell University