🔬 applied physics

Emergent Non-Hermitian Topology in Multi-Robot Network

This paper experimentally demonstrates the realization of programmable non-Hermitian topological phases in decentralized multi-robot networks, where digitally engineered non-reciprocal interactions induce emergent topological zero modes and skin effects that can be dynamically morphed across one to three dimensions.

Original authors: Jielong Zhang, Guiju Duan, Tinggui Chen, Shengjie Zheng, Bozheng Xue, Baizhan Xia

Published 2026-06-11

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Jielong Zhang, Guiju Duan, Tinggui Chen, Shengjie Zheng, Bozheng Xue, Baizhan Xia

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a swarm of tiny, dancing robots. Usually, when we think of robots working together, we imagine them following a strict leader or a central computer telling them exactly what to do. But in this research, the scientists created a system where the robots figure things out on their own, just by talking to their immediate neighbors.

Here is the simple story of what they did and what they found, using everyday analogies:

The Setup: A Digital Dance Floor

The researchers built a network of robots that act like a giant, living grid.

The Physical Robots: These are real, spinning motors with sensors.
The Digital Robots: These are "ghost" robots. They don't have motors or wheels; they exist only as code inside a computer.
The Connection: The real robots and the ghost robots talk to each other instantly. A real robot tells a ghost robot where it is, and the ghost robot tells the next real robot where to go. Together, they form a massive, seamless chain that can stretch from 1D (a line) to 2D (a flat sheet) to 3D (a cube).

The Magic Trick: "One-Way" Rules

In the normal world, if you push a friend, they push back with equal force (Newton's Third Law). But in this experiment, the scientists programmed the robots to break this rule. They created "non-reciprocal" interactions.

Think of it like a game of "Red Light, Green Light" where the rules are rigged:

If Robot A pushes Robot B, Robot B might push back harder.
Or, Robot A might push Robot B, but Robot B doesn't push back at all.

By tweaking these "one-way" rules, the scientists could make the entire group of robots behave in surprising, topological ways.

The Two Main Dances They Discovered

1. The "Zero Mode" Dance (The Perfect Wave)
When the robots are tuned to a specific rhythm (a specific frequency), they can perform a special dance called a Topological Zero Mode (TZM).

The "Delocalized" Version: Imagine a wave of energy that travels through the entire line of robots. Every robot spins with the exact same strength, passing the energy down the line like a perfectly synchronized conga line. It doesn't matter how long the line is; the energy flows smoothly everywhere.
The "Localized" Version: By changing the "one-way" rules, the scientists could suddenly stop the wave. The energy would get stuck at one end of the line, like a crowd of people all huddling in a corner while the rest of the room stands still.
The Cool Part: They could switch between these two states instantly, just by changing a number in the computer code. They could make the energy flow freely or get trapped in a corner at will.

2. The "Skin Effect" (The Crowd Surge)
When they changed the music to a slightly different, off-key rhythm, something else happened. Instead of a smooth wave, all the energy suddenly crashed into the very edges of the group.

Imagine a crowd of people in a room. If you shout a specific command, everyone in the middle stays calm, but everyone near the walls starts jumping wildly.
In the robot network, the "skin" of the group (the outer edges) absorbed all the energy, while the "insides" remained quiet. This is called the Non-Hermitian Skin Effect.

Why This Matters (According to the Paper)

The paper claims this is a big deal because:

It's Not Just Physics: Usually, these weird "skin effect" and "zero mode" behaviors are only seen in light waves, sound waves, or tiny atoms. This is the first time they have been seen in a robot network.
No Boss Needed: The robots didn't need a central commander to coordinate this. The complex, robust behavior emerged naturally from simple, local conversations between neighbors.
Digital + Physical: They proved that mixing real robots with "ghost" digital robots makes the system much more stable and precise. The digital robots act like a safety net, keeping the real robots from getting confused or making mistakes as the chain gets longer.

The Bottom Line

The researchers showed that you can turn a group of simple, decentralized robots into a programmable "topological machine." By just changing the rules of how they talk to each other, you can make the whole group act like a single, flowing wave, or suddenly collapse into a tight knot at the edge. It's a new way to understand how complex, robust groups can form without a leader, using the strange math of "non-Hermitian" physics.

Technical Summary: Emergent Non-Hermitian Topology in Multi-Robot Network

Problem Statement
Non-Hermitian (NH) topology, characterized by complex eigenvalue windings, exceptional points, and phenomena such as NH skin effects (NHSEs) and topological zero modes (TZMs), has been extensively studied in wave and matter systems (photonic, acoustic, mechanical, etc.). However, a fundamental constraint limits current implementations: these systems typically rely on physical linkages or global wavefield coherence within passive or semi-passive media. While active metamaterials can synthesize non-reciprocal feedback, their couplings remain restricted to passive elastic components. Conversely, biological swarms exhibit complex collective behaviors emerging from local, real-time state exchanges without centralized coordination. It remains an open question whether NH topological phases can emerge in collective intelligence systems that are unconstrained by spatial dimensionality and free from physical mechanical couplings.

Methodology
The authors experimentally realize programmable NH topological phases within decentralized multi-robot networks spanning one to three dimensions. The core methodology involves:

Decentralized Architecture: A network of active robots (physical agents) and digital counterparts (virtual agents) operates without a global coordinator. Each robot autonomously regulates its dynamics based on programmed coupling rules, its own kinematic state, and real-time feedback from neighbors.
Digital Programming of Non-Reciprocity: Instead of physical springs or interconnects, non-reciprocal interactions are implemented via digital control protocols. Robots exchange angular displacement information ( $\theta$ $θ$ ) to generate torques.
- Non-reciprocal hopping: Torques are generated via linear angular dependence with a non-reciprocal constant $\delta$ (e.g., $M_{2a-1} = -\delta \theta_{2a}$ and $M_{2a} = \delta \theta_{2a-1}$ ).
- Reciprocal interactions: Modeled as spring-equivalent stiffness coefficients ( $k_1, k_2$ ) producing reciprocal torques.
Hybrid Physical-Digital System (HMRN): To mitigate mechanical damping and error accumulation, the system employs a hybrid architecture where physical robots interface with digital robots via bidirectional communication. Physical robots execute real rotations, while digital robots perform virtual rotations, extending the lattice length and enhancing convergence to theoretical models.
Dimensional Scaling: The framework is demonstrated in:
- 1D: A Su-Schrieffer-Heeger (SSH) model with 11 physical and 90 digital robots.
- 2D: A square lattice with flux quantization ( $\pi$ ) comprising 51×51 robots (5×5 physical, 2576 digital).
- 3D: A cubic lattice with 15×15×15 robots (3×3×3 physical, 3348 digital).
Excitation and Control: The system is driven by periodic torques at specific eigenfrequencies ( $f_{TZM}$ ) to excite TZMs. By deviating the driving frequency from the eigenfrequency or tuning non-reciprocal parameters ( $\delta$ ), the system transitions to NH skin effects.

Key Contributions

Experimental Realization: The first experimental demonstration of programmable NH topological phases in a decentralized multi-robot network, extending from 1D to 3D.
Decoupling from Physical Linkages: The work establishes that NH topological states can emerge from localized interaction dynamics and real-time state exchange, bypassing the need for physical mechanical couplings (springs/beams) or global wave coherence.
Programmable Morphing: The ability to dynamically tailor non-reciprocal parameters ( $\delta$ ) to precisely morph TZMs between localized and delocalized states across different dimensionalities.
Hybrid Robustness: The demonstration that integrating digital robots into the network significantly reduces error accumulation and enhances the robustness of physical robot dynamics compared to purely physical networks.

Results

1D Network:
- Delocalized TZMs: Observed when $\delta = 0.2$ N·m/rad, characterized by uniform angular displacements and $\pi$ -phase alternations between adjacent odd-indexed robots.
- Localized TZMs: Observed when $\delta = -0.2$ N·m/rad, showing exponential decay of amplitude.
- NH Skin Effects (NHSEs): Triggered by off-resonant excitation (0.53 Hz), resulting in energy localization at the right boundary ( $\delta > 0$ ) or left boundary ( $\delta < 0$ ).
- Error Mitigation: Hybrid networks (HMRN) showed significantly higher convergence to theoretical amplitude ratios ( $\alpha$ ) compared to purely physical networks (PMRN), where errors accumulated with distance.
2D Network:
- State Control: By tuning $\delta_x$ and $\delta_y$ , the system achieved delocalized bulk modes, edge-localized modes (along one axis), and corner-localized modes.
- NHSEs: Off-resonant driving caused a collapse of TZMs into corner-localized NH skin modes.
- Physical Limitations: Purely physical 2D networks exhibited degraded control precision and stabilization compared to hybrid systems.
3D Network:
- 3D Confinement: Successfully demonstrated delocalized, surface-localized, edge-localized, and corner-localized TZMs by adjusting $\delta_x, \delta_y, \delta_z$ .
- Transition to Skin Modes: Deviating the driving frequency (0.62 Hz) from the eigenfrequency (0.3 Hz) caused all TZM configurations to collapse into 3D corner-localized NH skin modes.

Significance
The paper claims to establish multi-robot networks as a highly reconfigurable and scalable platform for exploring non-equilibrium topological physics. By decoupling NH topology from traditional wave-based requirements, the work provides a synthetic counterpart to natural self-organizing systems (like biological swarms). The significance lies in:

New Paradigm for Active Matter: Offering a framework for topologically protected, robust collective behaviors in active matter that are adaptable to dynamic environments.
Dimensional Flexibility: Unlocking the exploration of high-dimensional NH physics without the constraints of physical lattice construction.
Bridge to Quantum Principles: Bridging quantum-like topological principles with macroscopic active systems, potentially accelerating the discovery of exotic out-of-equilibrium topological phases.

The authors emphasize that this decentralized mechanism, previously unattainable in traditional physical platforms, redefines the design space for non-Hermitian topology.