The study, led by Professor Yutaka Sumino and Assistant Professor Kiwamu Yoshii of the Tokyo University of Science, focuses on a phenomenon known as nonreciprocal interaction. In a conventional passive system, interactions generally obey action–reaction symmetry: if one particle exerts a force on another, the second particle responds with an equal and opposite force. Nonreciprocal interactions violate this balance. One particle can influence another more strongly than it is influenced in return, creating behavior that is impossible for isolated passive particles in equilibrium.

To build this unusual system, the researchers suspended polystyrene colloids in water between transparent electrodes coated with indium tin oxide. The suspension contained two particle sizes, with radii of approximately 1 and 1.5 micrometers. When the team applied an alternating electric field, fluid began to circulate around the particles through electrohydrodynamic, or EHD, effects. These flows were not simply a passive response to the field: they were generated around each particle and altered the motion of neighboring particles throughout the suspension.

The crucial feature was the difference in particle size. EHD flow strength depends strongly on particle dimensions, so the larger colloids produced substantially stronger flows than the smaller ones. This made the attraction between unlike particles unequal. A large particle could draw a small particle toward it more effectively than the small particle could influence the large one in return. The resulting imbalance produced a form of effective attraction that did not satisfy Newtonian action–reaction symmetry, even though the particles themselves were not equipped with motors or other self-propulsion mechanisms.

When differently sized particles came into contact, they spontaneously formed asymmetric pairs. Each pair had a distinct front and back, rather than the symmetric shape expected from ordinary aggregation. Because the surrounding EHD flows acted differently on the two particles, the pair began to move through the liquid as a self-propelled unit. In effect, one particle “chased” the other, and the coupled structure translated through the suspension. The motion emerged from the interaction between the particles and the driven fluid, not from any internal propulsion system.

As thousands of these active pairs formed, they gathered into larger clusters. Yet the clusters did not behave like ordinary colloidal aggregates, which typically grow as particles continue to attach. Instead, the groups remained dynamic. They repeatedly fragmented, rearranged their internal structure, and reassembled. This process, known as arrested coarsening, prevented the suspension from undergoing unlimited aggregation. The particles continued to attract one another, but the motion generated by the nonreciprocal pairs disrupted the growth of giant, static clumps.

The contrast with a single-size suspension was particularly revealing. When the researchers used particles of only one size, the interactions remained reciprocal because the particles generated comparable EHD flows. Those suspensions gradually developed static crystalline aggregates. The result showed that the persistent cluster dynamics were not caused simply by the electric field or by attractive forces alone. Instead, the combination of unequal particle size, asymmetric EHD interactions, and pair propulsion was essential for producing the continuously evolving structures.

Numerical simulations supported the experimental findings and indicated that nonreciprocal propulsion at the level of individual pairs is the minimal ingredient needed to reproduce the persistent cluster motion. The simulations showed how moving pairs can inject activity into otherwise passive assemblies, driving internal rearrangements and repeatedly interrupting the normal coarsening process. Because the same principles can apply to other systems with unequal responses, the researchers suggest that comparable dynamics may occur in biological collectives, including cell colonies and animal groups, where agents can exert different influences on one another.

The findings offer an experimentally controllable platform for studying nonequilibrium many-body physics on a scale far larger than the small clusters examined in many earlier experiments. They also point toward a broader design principle for programmable materials: by engineering asymmetric interactions rather than giving every component an independent motor, researchers may be able to create materials that assemble, move, divide, and reorganize under external control. Such systems could eventually inspire microrobotic swarms, adaptive materials, and artificial collectives that remain dynamic instead of locking into a fixed structure. As Professor Sumino explains, the work shows how breaking action–reaction symmetry can allow matter to form dynamic order spontaneously—attracting, gathering, and then splitting instead of simply becoming an ever-larger clump.

Subject of Research: Not applicable
Article Title: Arrested coarsening in active colloidal suspensions driven by nonreciprocal electrohydrodynamic interactions
News Publication Date: 6-Aug-2026
Web References: https://doi.org/10.1103/96ky-d1p9
References: Physical Review Letters, “Arrested coarsening in active colloidal suspensions driven by nonreciprocal electrohydrodynamic interactions,” DOI: 10.1103/96ky-d1p9
Image Credits: Professor Yutaka Sumino, Tokyo University of Science, Japan

Keywords

Active matter, nonreciprocal interactions, colloidal particles, electrohydrodynamics, self-propelled particle pairs, arrested coarsening, nonequilibrium physics, collective motion, programmable materials, microrobotics

The fundamental principle governing link-bots is rooted in the concept of emergent functional dynamics, whereby simple units, when coupled with physical constraints, give rise to complex, adaptive behaviors. Unlike conventional swarm robotics, which typically relies on energy-intensive sensors, communication devices, and onboard computation to coordinate movements and actions, link-bots harness geometry and mechanical interactions. Each particle within the chain possesses legs oriented at an angle, which interact with a uniformly vibrating surface to induce self-propulsion. This design negates the need for internal power sources, enabling energy-efficient, spontaneous locomotion that surprises even seasoned roboticists for its elegance and simplicity.

Harvard’s L. Mahadevan, a distinguished scholar bridging applied mathematics, physics, and evolutionary biology, co-led this study, highlighting the interdisciplinary approach that made this achievement possible. Collaborating with Professor Ho-Young Kim from Seoul National University, the team moved beyond traditional robotic paradigms to embrace principles widely observed in natural collective systems. Their publication, slated for release in Science Advances, meticulously details the experimental results, computational modeling, and the underlying physics that enable these chains of particles to exhibit life-like coordinated behaviors without centralized commands.

The emergent behavior of these link-bots is astonishingly versatile. By adjusting the architecture of the links, the chain ensembles can modulate their movement patterns—accelerating, stopping, reversing, or squeezing through tight spaces with remarkable dexterity. This adaptability extends beyond mere locomotion; link-bots can physically interact with objects, collectively surrounding and transporting them, overcoming challenges that a single unit could not surmount. Such collective adaptability arises from simple mechanical interactions rather than complex sensory inputs, which paves the way for low-power solutions in fields requiring coordination in constrained or unpredictable environments.

To parse the intricate dynamics of these robotic collectives, the team employed advanced computational models, spearheaded by postdoctoral fellow Kimberly Bowal. These simulations explore how variations in link configurations and the number of particles affect overall motion and behavior. The modeling has been invaluable in probing scenarios difficult to test empirically and offers predictive power for engineering new functionalities. Bowal emphasizes that the programmable behaviors emerge purely from physical linkage and environmental feedback, showcasing a paradigm where robotics intelligence is distributed across geometry and interaction patterns rather than encoded centrally.

This shift in outlook stands in stark contrast with traditional top-down designs where every trajectory, task, or response is pre-planned and enforced by onboard intelligence. Instead, the link-bots exemplify a bottom-up approach, where collective organization and emergent functionality arise spontaneously from simple locally governed interactions. Mahadevan reflects on this fundamental departure, proposing that the principles elucidated by their work mirror biological evolution’s indifference to planners, relying on the inherent power of self-organization to generate function and complexity.

From a technical perspective, the physical construction of link-bots leverages mechanical engineering concepts including modularity, compliant mechanisms, and vibrational energy conversion. The notched links act as flexible joints, permitting both connectivity and nuanced relative motion among particles. The tilted legs of each module translate ambient vibrations into forward thrust, a physical phenomenon manifesting as rectification of oscillatory motion—a concept well-studied in physics but innovatively applied here to microrobotics. This careful orchestration of mechanical design principles culminates in a system where the whole truly exceeds the sum of its parts.

Furthermore, the implications for applications in multiple domains are profound. Potential uses could range from micro-scale transport systems capable of autonomously sorting and conveying objects, to adaptive structures that change shape and function on demand. Since these robots operate without conventional power sources, they hold promise for deployment in delicate environments or in scenarios where recharging or maintenance is impractical. The simplicity of their design also suggests scalability, with swarms potentially numbering in the hundreds or thousands, cooperatively tackling tasks that require both flexibility and resilience.

The research also touches on fundamental questions about the nature of intelligence and control in engineered systems. By demonstrating how complex behaviors can arise absent centralized planning—through geometry and local coupling—this study challenges prevailing dogmas in robotics and computational science. It invites a reconsideration of how future robotic collectives might be designed, leveraging physical principles as integral components of their “programming.” This could herald a novel era where robotics blurs the boundaries between the mechanical and the biological, embodying concepts from evolutionary biology within synthetic constructs.

Mahadevan and his collaborators are optimistic that this work represents but the initial foray into a wider domain of robot collectives governed by emergent physical interactions. The continued blending of mathematics, mechanical engineering, and biology promises to unlock new classes of devices that rethink autonomy and adaptability. As these systems evolve, they might illuminate long-standing mysteries in both robotics and nature regarding how cooperation and complexity arise from simplicity.

The scientific community eagerly awaits the full release of their paper in Science Advances on May 9, 2025, which promises to provide comprehensive experimental data and theoretical models underpinning these findings. The partnership between Harvard SEAS and Seoul National University exemplifies the power of international collaboration in pushing the frontiers of knowledge and technology. The link-bot project not only advances robotic science but also underscores the elegance and utility of nature-inspired design philosophies.

In a broader context, the link-bots demonstrate the potential for a paradigm shift—from engineered systems meticulously controlled by humans to self-organized, self-sufficient robotic collectives. These collectives capitalize on the physics of interactions rather than the metaphysics of programming. The team’s approach may inspire future generations of roboticists to embrace minimalism and physicality, opening new pathways for innovation in swarm robotics and beyond.

As research continues, one can envision these link-bots paving the way for transformative advances in soft robotics, microrobotics, and applied physics, where intelligence is emergent, collective, and embedded in the very fabric of their construction. With such systems, the boundary between machine and organism becomes intriguingly blurred, hinting at a future where robotic swarms operate with the grace and efficiency of biological systems—self-organized, resilient, and profoundly adaptive.

Subject of Research: Not applicable

Article Title: Emergent functional dynamics of link-bots

News Publication Date: 9-May-2025

Web References:
https://www.science.org/doi/10.1126/sciadv.adu8326

References:
Mahadevan, L., Kim, H.-Y., Son, K., Kim, K. (2025). Emergent functional dynamics of link-bots. Science Advances, DOI: 10.1126/sciadv.adu8326.

Image Credits: Mahadevan Lab / Harvard SEAS

Keywords: Soft robotics, Artificial intelligence, Robotic designs, Robots, Microrobots, Applied mathematics, Algorithms, Computational science, Mathematical modeling, Mathematics, Physics, Applied physics, Mechanical engineering, Mechanical components