Source-linked AI summary
Swarming, swirling and stasis in sequestered bristle-bots
L. Giomi, N. Hawley-Weld, L. Mahadevan
TL;DR
The paper investigates how confinement and boundaries shape transitions between disordered and collective motion. It studies self-propelled Bristle-Bots and simulations, finding density-dependent swirling, jamming, and random-motion regimes whose occurrence depends on agents’ angular motion.
Problem
Collective behavior is often studied while minimizing or omitting boundaries, leaving their role in transitions between disordered and organized motion less examined.
Source: §2 · p. 4, §4 · p. 11Method
The authors combine experiments on tunable self-propelled Bristle-Bots with simulations of confined ellipses and polar disks.
Source: §1 · p. 4, §4 · p. 9, §5 · pp. 12–13Results
Polar-disk simulations show random motion at low density, boundary jamming at high density with zero angular velocity, and boundary-circulating clusters at high density with finite angular velocity.
Source: §5 · p. 13Takeaways & Limitations
Coordinated circulation and jamming in confined agents are suggested to be generic, while cluster structure depends on particle shape and packing.
Source: §5 · p. 13
Abstract
from arXiv · showhide
The collective ability of organisms to move coherently in space and time is ubiquitous in any group of autonomous agents that can move and sense each other and the environment. Here we investigate the origin of collective motion and its loss using macroscopic self-propelled Bristle-Bots, simple automata made from a toothbrush and powered by an onboard cell phone vibrator-motor, that can sense each other through shape-dependent local interactions, and can also sense the environment non-locally via the effects of confinement and substrate topography. We show that when Bristle-Bots are confined to a limited arena with a soft boundary, increasing the density drives a transition from a disordered and uncoordinated motion to organized collective motion either as a swirling cluster or a collective dynamical stasis. This transition is regulated by a single parameter, the relative magnitude of spinning and walking in a single automaton. We explain this using quantitative experiments and simulations that emphasize the role of the agent shape, environment, and confinement via boundaries. Our study shows how the behavioral repertoire of these physically interacting automatons controlled by one parameter translates into the mechanical intelligence of swarms.
1. Motion of an individual BBot
A Bristle-Bot’s eccentric motor flexes tilted toothbrush bristles, whose asymmetric slipping produces forward motion. Adjusting bristle geometry and battery position tunes its speed and balance of walking and spinning.
1. Motion of an individual BBot
- Sliding the battery adjusts the center of mass and the relative amounts of translational and rotational speed.The design also uses an elliptical chassis, a cell-phone motor, and removable bristle wedges to vary bristle inclination.Source: §1 · p. 3
- The eccentric motor periodically loads and unloads tilted bristles, whose slipping against the substrate produces net forward motion.The bending and unbending rectify the motor’s periodic drive into displacement.Source: §1 · p. 3, Figure 2 · p. 3
- Increasing bristle angle from 5° to 30° substantially slows BBots, while longer bristles increase speed and favor straighter paths over short-bristle circling.Short, stiff bristles also produce noisier motion with rebounds and jumps.Source: §1 · pp. 3–4, Figure 3 · p. 4
- With 5 mm bristles, spinners at α = 0° or 5° turn clockwise, while walkers at α ≥ 10° travel straight or on weakly curved paths.Spinners can reach angular velocities of up to 30 rad/s while moving slowly.Source: §1 · p. 4
2. The effect of boundaries and topography
Boundary shape changes how individual Bristle-Bots move at the arena edge. A soft boundary redirects them inward or produces walker oscillations, whereas a hard wall aligns them to travel along its edge.
2. The effect of boundaries and topography
- The arena is a 44 cm circular plate, and the authors note that boundaries are often minimized or omitted in prior collective-behavior studies.A single Bristle-Bot occupies approximately 0.8% of the available area.Source: §2 · p. 4
(a) Soft boundaries
- At a soft boundary, spinners turn back toward the arena’s interior, while walkers typically oscillate near the edge before returning inward.Strong walkers with α = 20° avoid these oscillations because forward propulsion dominates sideways spinning.Source: (a) Soft boundaries · p. 5, Figure 4 · p. 5
(b) Hard boundaries
- A hard vertical wall aligns bots parallel to the edge, after which they travel around the arena clockwise or counterclockwise depending on their initial contact angle.Unlike the soft boundary, the hard boundary does not reflect bots inward.Source: Figure 4 · p. 5, (b) Hard boundaries · p. 6
3. Collective behavior of BBots - Experiments
In a soft-boundary arena, increasing population yields different collective states for spinners and walkers: swirling clusters or stasis. The experiments also reveal hysteresis and cluster behavior shaped by boundary type and bristle angle.
3. Collective behavior of BBots - Experiments
- The experiments track bot positions, orientations, and translational and rotational velocities from backlit video recorded at 40 fps.The experimenters released bots from an initial template or changed population from 2 to 16 to examine state transitions and hysteresis.Source: §3 · p. 6
- The average velocity and average speed distinguish disordered, coherent, and arrested motion: disordered groups have near-zero average velocity, while arrested clusters have both measures near zero.The center-of-mass trajectory provides an additional view of the collective state.Source: §3 · p. 6, Figure 5 · p. 7
- With soft boundaries, spinners are disordered below 10 bots but form a coherent swirling cluster above 10, while walkers form stable stationary clusters above 8.The figure contrasts low-density spinners, high-density spinners, and high-density walkers; the respective examples show 7, 24, and 15 bots.Source: §3 · pp. 6–7, Figure 5 · p. 7
- Once formed, a swirling spinner cluster persists as bots are withdrawn until its population falls below 6.The phase diagram summarizes disordered motion, swirling, and stasis against population and bristle angle.Source: §3 · p. 7, Figure 6 · p. 8
- At intermediate inclination, clusters of four or five bots can remain stable for over a minute before coordinated oscillations disassemble them; denser clusters persist much longer.The authors describe the high-density stasis region as clusters stable over very long times.Source: §3 · p. 8
- At a hard boundary, bots sliding along the wall form groups that can abruptly self-arrest as group size increases, producing half-aster shapes.These differ from the orthogonally oriented jammed structures formed with soft boundaries.Source: §3 · p. 8
4. Collective behavior of BBots - Theory
The theory models Bristle-Bots as self-propelled ellipses whose contact interactions and boundary torques shape collective behavior. Simulations reproduce disordered motion, swirling, and stasis, with outcomes depending on density, spinning, and confinement.
4. Collective behavior of BBots - Theory
- Ellipses self-propel along their major axes, repel overlapping neighbors elastically, align through contact torques, and receive an inward torque from the boundary.The boundary torque is non-local and exponentially decaying; its range is tuned by a length scale.Source: §4 · pp. 9–10
- Curvature and longer-range boundary interactions pack particles more densely at the edge, demonstrating how the physical environment can alter collective behavior.Extending the interaction range accentuates the boundary's focusing effect.Source: §4 · p. 10
- The model uses four dimensionless parameters, while density and spinning-to-walking ratio are the relevant experimental variables because orienting strength and noise are intrinsic to the agents.The model also distinguishes itself by using torque balance and non-local interactions to represent orientation and boundaries.Source: §4 · pp. 10–11
- At high density, walkers can become trapped at the boundary, while spinners tilt and circulate the cluster; increasing angular velocity lets walkers push neighbors aside and escape.The number of bots required for jamming decreases with angular velocity.Source: §4 · p. 11
- Simulations reproduce random motion at low density, swirling by spinners and stasis by walkers at high density; without confinement, the ellipses show flocking instead.The simulated regimes match the observed behaviors and depend on the interplay of self-propulsion, particle geometry, and confinement.Source: §4 · p. 11, Figure 9 · p. 12
5. Discussion
Simulations with ellipses and polar disks show three density- and rotation-dependent regimes, supporting the possibility that circulation and jamming are generic in confined agents. Collective motion is described as robust to particle shape even as cluster structure depends on shape and packing, with possible biological and exploratory applications.
5. Discussion
- Polar-disk simulations show random motion at low density, boundary jamming at high density with zero angular velocity, and boundary-circulating clusters at high density with finite angular velocity.The study reports these regimes as analogous to those of the ellipses, despite replacing physical torque with a generic aligning interaction.Source: §5 · pp. 12–13
- The similarity across experiments and models suggests that circulation and jamming are generic in confined agents, although cluster structure depends on particle shape and packing.The authors describe collective motion as robust and dependent on simple, non-specific principles.Source: §5 · p. 13
- The authors propose testing confinement and substrate effects in termite swarms and using BBots to sense roughness or search curvature as leaderless explorers.They suggest that spinner and walker BBots could explore environments in parallel.Source: §5 · p. 13
Appendix A. Locomotion of an individual BBot
The appendix models a single BBot’s gait as quasistatic bristle bending and slipping under cyclic eccentric loading. It derives bristle shapes and per-cycle step size, while noting the model’s restriction for light bots that lose contact with the substrate.
Appendix A. Locomotion of an individual BBot
- Each motor cycle loads the bristles, then reduces the load so they unbend and slip on the substrate, producing a forward step.The bristles’ forward displacement is linked to the sequence of loading, recoil, and slip.Source: Appendix A · p. 14
- The model treats the bristles as an ideal elastic beam in quasistatic equilibrium, with force and torque balance and an external tip force; it uses µ = 0.1.The schematic describes a time-dependent vertical load and kinetic friction along the horizontal direction.Source: Appendix A · pp. 14–16, Figure 11 · p. 15
- The beam’s constitutive relation yields an angle equation with cantilever boundary conditions, which the appendix solves using Jacobi elliptic functions.The solution is used to construct the shape of the deflected bristles.Source: Appendix A · pp. 15–16
- The model obtains per-cycle step size from the bristle tip’s displacement between maximum and minimum load, using the calculated bristle shape.The analysis focuses on horizontal displacement and assumes quasistatic bristle conformation.Source: Appendix A · pp. 14, 16, 17, 18
- The gait model captures qualitative dependencies consistent with experiments, but light BBots that lose substrate contact and jump fall outside its tractable gait analysis.The jumping motion couples to bristle dynamics, and the appendix states that its methods cannot handle that gait.Source: Appendix A · pp. 14, 18
Appendix B. Collective behavior of self-propelled disks
The disk simulations test whether coordinated behaviors persist beyond the Bristle-Bots’ shape-dependent interactions. They model isotropic self-propelled disks in a circular arena and exhibit robust coordinated regimes, alongside shape-sensitive cluster structure and additional dynamical modes.
Appendix B. Collective behavior of self-propelled disks
- The isotropic disk comparison assigns Mij a short-range phase-synchronizing alignment rule without a direct physical origin.The particles have center-of-mass positions and orientations, and are confined to a circular domain whose boundary interaction matches the elliptical-particle simulations.Source: Appendix B · p. 18
- In a 15-disk simulation, treadmilling carries a tracer from the cluster’s leading end to its trailing end, then back toward the front along the interior.The tracer remains near the trailing end for about one loop before progressing forward.Source: Figure 14 · p. 19
- Swarming, swirling, and stasis remain robust across confined self-propelled-agent systems, although cluster structure and packing depend on system details, especially particle shape.The reported robustness is also insensitive to inertia and to the precise form of the aligning torque.Source: Appendix B · p. 19
- In a 300-disk simulation, breathing alternates dense central packing with sudden bursts that move particles toward the boundary.The figure caption links dips in v2 to central packing and peaks in v1 to bursts of boundary migration.Source: Figure 15 · p. 19, Appendix B · p. 20