Multimodal collective swimming of magnetically articulated modular nanocomposite robots

Collective behavior in active matter spans from cells to animal swarms and inspires robotic collectives capable of complex tasks. In miniaturized robotics, external-field-driven swarms avoid on-board power but most prior collective systems involve colloidal particles, which face challenges in transporting many cargos rapidly and with positional control. Polymer-composite robots enable diverse geometries for manipulation, yet transportation has largely focused on single small cargos and remains limited for simultaneous, rapid delivery of many objects. This study aims to achieve controllable collective actuation of multiple miniaturized robots with on-demand multimodal swimming to generate and regulate vortices at the air–water interface for transporting multitudinous cargos. The authors introduce musculoskeletal-mimicking nanocomposite robots that are stiff yet lightweight, capable of agile above-water swimming, and magnetically modular, to overcome limited interactive motions in multi-robot polymeric systems.

The paper situates its work within nature-inspired and engineered collectives, including kilobot and drone swarms, and externally driven microswarms actuated by magnetic, optical, electric, or acoustic fields. Time-varying magnetic fields enable programmable microrobot motility via magnetic torque and have produced reconfigurable swarms of magnetic microparticles. However, collective cargo manipulation by particle swarms struggles with transporting multiple cargos over distance and with precise control. Prior polymeric microrobots with varied shapes can load and release small cargos, but typically handle single cargos smaller than the robots. Although multiple robots have cooperatively carried larger, heavier single objects, achieving rapid, simultaneous delivery of numerous cargos remains challenging, motivating controllable multimodal collective actuation of modular robots.

Design and materials: Robots mimic a musculoskeletal hierarchy using a nanoporous carbon nanotube yarn (CNTY) framework (stiff, lightweight) coated with a magnetic polymer composite (PDMS binder plus iron particles) localized at the surface. High aspect ratios (AR 2.5, 6, 13; up to 21.5 in capability) were targeted to enhance hydrodynamic volume when magnetically assembled.

• CNTY synthesis: Multi-walled carbon nanotube (MWNT) fibers were synthesized via floating catalyst chemical vapor deposition at 1200 °C using methane (carbon source), thiophene (promoter), and ferrocene (catalyst precursor). Fibers were drawn through water and acetone baths, collected on a roller (5 m min−1). One hundred MWNT fibers were twisted into a ~340 μm-thick yarn with porous interior. MWNTs were ~10-walled, ~14 nm thick; the yarn provided specific strength 1.19 N tex−1 and specific modulus 46.9 N tex−1.
• Dip-coating and curing: PDMS (Sylgard 184, 10:1 prepolymer:crosslinker) with dispersed iron particles (0–40 vol%) was degassed. Six-centimeter CNTY specimens were dip-coated horizontally for 1 min, then drained vertically for 10 min at room temperature and cured at 130 °C for 3 h. The nanoporous CNTY allowed PDMS infiltration by capillarity while 5 μm iron particles remained near the surface, forming a core–shell-like architecture. After curing, robots were cut to ARs 2.5, 6, 13.
• Rheology and magnetic properties: Complex viscosity of neat PDMS prepolymer was 18 Pa·s at 3 rad·s−1. With 10, 20, 30 vol% particles, complex viscosities were ~90, 371, 897 Pa·s, yielding magnetic layer thicknesses of ~14, 21, 60 μm, respectively. Saturation magnetization (Ms) increased with loading: ~69, 118, 152 emu·g−1 for 10, 20, 30 vol%. Above 30 vol% the viscosity hindered uniform coating. Magnetic actuation setup and analysis: Robots swam at the air–water interface under a pulsed quadrupolar electromagnetic field (max ~13 mT). Two orthogonal coil pairs alternated on/off to produce a clockwise rotating field. Magnetic frequency was varied (e.g., 3.3, 6.7, 10, 13.3, 16.7, 20 Hz) to switch between modes. High-speed imaging tracked center-of-mass trajectories to classify modes via x–y harmonic analysis. Multiple robots were tested for magnetic assembly and collective behaviors. Vortex visualization used NIR illumination (0.5 W cm−2 for 3 min) and FLIR thermal imaging to map water surface temperature gradients. Cargo transport used seven-robot assemblies and a motorized 2-DoF stage coupled to the coils for precise steering. Microbead/microplastic counts were estimated from area coverage analysis of images with bead size and packing voids accounted for.

• Agile, multimodal above-water swimming: Robots exhibit on-demand switching between counterclockwise rectilinear translational swimming and clockwise rotational swimming by tuning magnetic frequency. Rotation of robots synchronized to magnetic source frequency.
• High speeds: An AR-2.5/30 vol% robot achieved average speed up to ~212 mm s−1 (≈180 body lengths s−1), about twice that of water striders and 47× faster than a prior binary-nanocomposite spinbot. AR-6/30 vol% and AR-13/30 vol% reached ~66 and ~40 BL s−1, respectively.
• Mode control parameters: Increasing frequency reduced centroid–vertex distance and promoted rotational mode; increasing particle concentration increased inertial forces and enabled longer AR operation. 30 vol% robots supported bimodal swimming up to AR ≈ 21.5.
• Collective modes and magnetic modular assembly: Multiple robots assembled via dipole–dipole interactions into chain-like configurations exhibiting three collective modes: assembled rectilinear translational swimming, assembled rotational swimming, and fluctuating rotational swimming (engagement–disengagement cycles when drag exceeded magnetic attraction). Assemblies could perform rectilinear swimming even when a single equivalent long robot (e.g., AR 21.5) was nonmotile due to drag.
• Synergistic motility: Co-actuation improved performance of ineffectual robots; e.g., a single AR-13/10 vol% robot only rotated, but achieved rectilinear swimming at 3.3 Hz when linked to AR-6/10 vol% and rotational swimming at 20 Hz when linked with AR-2.5/10 vol% and AR-6/10 vol%.
• Vortex generation and control: Assemblies generated Rankine vortices with controllable magnitude, angular velocity, and chirality via mode switching and frequency tuning. 10 vol% assemblies at 6.7 Hz produced stronger vortices than at 3.3 Hz. 30 vol% robots intensified vortices, forming a paraboloid water surface at 13.3 Hz, and enabled chiral inversion during reorganization between modes.
• Cargo transport via vortices: Seven 10 vol% robots transported 3,350 floating microbeads (D = 255 μm) into a confined space by switching from rectilinear (blocking) to rotational (dragging and conveying) modes. Seven 30 vol% robots transported four semi-submerged spherical cargos (D = 4 mm, 52 mg each; ~82× an individual robot’s mass) to a target via combined CW vortex-induced flow and direct pushing; assemblies reformed after transient disengagement. Precision improved using a 2-DoF motorized stage.
• Large-scale microplastics collection: Seven 30 vol% robots at 5 Hz gathered and merged dispersed clusters, collecting a total of 4,630 floating polyethylene microplastics (D = 850 μm) within 150 s in an open space.

The study demonstrates that musculoskeletal-mimicking CNTY-based nanocomposite robots can be magnetically articulated into modular assemblies that achieve on-demand multimodal collective swimming. This collective actuation addresses the challenge of transporting numerous small cargos by leveraging vortex generation and control at the air–water interface, rather than grip-based manipulation. The stiff yet lightweight CNTY core and surface-localized magnetic composite maximize magnetic interactions among modules while maintaining agility, enabling chain assemblies that overcome drag limitations of single long robots. Adjustable swimming modes and frequencies tailor vortex magnitude and chirality, facilitating tasks from blocking to transporting and confining thousands of particles. Intensifying the magnetic layer (30 vol%) further strengthens magnetic attraction and vorticity, allowing manipulation of heavier, semi-submerged cargos. These capabilities broaden the functional repertoire of carbon-based nanomaterial robots and suggest their utility in applications such as microplastic removal, microfluidic vortex control, and pharmaceutical transport, where rapid, collective handling of many small objects is required.

This work introduces ternary nanocomposite CNTY robots that, through magnetically articulated modular assembly, realize agile, above-water multimodal swimming and controllable vortex generation. Key contributions include synchronization-enabled high speeds (up to ~180 BL s−1), robust collective modes (assembled rectilinear, assembled rotational, fluctuating rotational), and vortex-induced transport enabling rapid handling of thousands of particles and heavy semi-submerged cargos. The approach leverages a biomimetic, stiff–lightweight architecture with surface-concentrated magnetic particles to enhance inter-robot attraction and assembly stability. The demonstrated capabilities point toward versatile applications in environmental remediation (microplastic collection), microfluidic flow control, and cargo transport in biomedical contexts. Future work can translate these principles to application-specific platforms (e.g., microfluidic devices or in situ remediation systems), further optimize module geometry and magnetic loading for scalability and robustness, and expand control strategies for more complex, autonomous multi-robot behaviors.

• Dependence on particle loading: Lower magnetic particle concentration (e.g., 10 vol%) led to weaker inter-module attraction; assemblies disassembled at comparatively lower frequencies and exhibited fluctuating rotational swimming due to drag overcoming magnetic forces.
• Mode/frequency constraints: Very long single robots (e.g., AR ~21.5) were nonmotile alone due to drag; collective assembly mitigates this, but operation windows remain bounded by drag–magnetic force balance.
• Interface specificity: Demonstrations primarily occurred at the air–water interface; performance and vortex-mediated transport may differ in other fluids, depths, or confined geometries.
• External field and staging: Operation requires an external pulsed electromagnetic setup (<13 mT) and, for precise steering, a motorized 2-DoF stage. This may limit portability and in situ deployment without specialized hardware.
• Single-robot limits: A single robot could not transport thousands of particles, highlighting reliance on collective assembly for high-throughput tasks.