pith. sign in

arxiv: 2511.00384 · v2 · pith:SRZLYJDSnew · submitted 2025-11-01 · ❄️ cond-mat.soft

Physics-informed digital twin and onboard control of a brainbot for intelligent active matter

Isa Mammadli , Prajol Shrestha , Jayant Pande , Filip Novkoski , Siddhant Mohapatra , Martial Noirhomme , Andreas Maier , Nicolas Vandewalle
show 1 more author
Ana-Suncana Smith
This is my paper

Pith reviewed 2026-05-22 11:41 UTC · model grok-4.3

classification ❄️ cond-mat.soft
keywords brainbotdigital twinmodel predictive controlactive matterbristlebotonboard controlkinematic modelautonomous trajectory tracking
0
0 comments X

The pith

A brainbot senses its state, predicts its motion, and computes control inputs onboard using a physics-informed digital twin.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper constructs a digital twin of a programmable bristlebot so the device can run its own model predictive control. A kinematic model is fitted to measured trajectories and then used to generate realistic future paths. This model lets the brainbot choose actions in real time to follow a chosen route. If the approach holds, active particles stop being objects that move according to fixed rules and instead become entities that adjust their own behavior. The result is a concrete step toward active matter that behaves like a collection of simple agents rather than passive particles.

Core claim

We address this challenge by realizing an autonomous brainbot, building on a recently developed programmable bristlebot. First, we construct a physics-informed digital twin of the device, based on a kinematic model that reproduces measured trajectory statistics and generates long, statistically faithful synthetic trajectories. The kinematics forms the foundation for implementing onboard model predictive control (MPC), enabling autonomous trajectory tracking, demonstrated by accurate execution of a non-trivial target path.

What carries the argument

The physics-informed digital twin built on a kinematic model fitted to trajectory statistics, which supplies the predictions needed for onboard model predictive control.

If this is right

  • The brainbot performs accurate autonomous trajectory tracking of non-trivial target paths.
  • Physical modeling, data-driven fitting, and control are combined in one onboard framework.
  • The platform supports machine-learning-enabled studies with multiple interacting agents.
  • The work establishes a foundation for creating intelligent, adaptive active matter.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the kinematic approximation remains accurate at higher speeds or in crowded settings, the same digital-twin method could support swarms that coordinate without external oversight.
  • The approach could be tested by replacing the bristlebot with other active-particle platforms that have onboard computation to see how far the kinematic assumption travels.
  • Adding simple learning rules on top of the current predictive control would let each brainbot improve its own model over time from its own trajectory data.

Load-bearing premise

The kinematic model, once fitted to measured trajectory statistics, generates synthetic trajectories faithful enough that the resulting model predictive control works accurately when run on the physical brainbot.

What would settle it

Execute the onboard model predictive control on the physical device and check whether the brainbot deviates substantially from the target path or whether the synthetic trajectories fail to match the statistics of real runs.

Figures

Figures reproduced from arXiv: 2511.00384 by Ana-Suncana Smith, Andreas Maier, Filip Novkoski, Isa Mammadli, Jayant Pande, Martial Noirhomme, Nicolas Vandewalle, Prajol Shrestha, Siddhant Mohapatra.

Figure 1
Figure 1. Figure 1: FIG. 1. (a) Position vector [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Properties of the model. Top row: Deterministic trajectories of the geometric center (shown in red) and the instanta [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Results of simulation (in red) compared to experimental data (in black). The upper row shows the trajectories, while [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 2
Figure 2. Figure 2: While the linear and circular (spinning and or [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. The top row shows synthetic trajectories similar to the four experimental trajectories shown in Fig. 2. The second [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 3
Figure 3. Figure 3: We first generate a histogram for the ω values for each trajectory across the entire run of the experiment (gray bars in [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. The top row shows synthetic trajectories corresponding to the experimental trajectory in the third column of Fig. 3, [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
read the original abstract

Establishing adaptive particles that sense their state, anticipate their evolution, and compute control inputs onboard has been a major challenge in non-equilibrium physics. We address this challenge by realizing an autonomous brainbot, building on a recently developed programmable bristlebot. First, we construct a physics-informed digital twin of the device, based on a kinematic model that reproduces measured trajectory statistics and generates long, statistically faithful synthetic trajectories. The kinematics forms the foundation for implementing onboard model predictive control (MPC), enabling autonomous trajectory tracking, demonstrated by accurate execution of a non-trivial target path. This provides a proof of principle for a brainbot that senses its state, predicts its evolution, and computes control inputs onboard, unlike conventional active particles with fixed motility, thereby transforming the brainbot into an agentic physical entity. By integrating physical modeling, data-driven parameter identification, and control into a unified framework, our approach provides a scalable platform for machine-learning-enabled multi-agent studies and lays the groundwork for intelligent, adaptive active matter.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript claims to realize an autonomous brainbot by constructing a physics-informed digital twin of a programmable bristlebot. A kinematic model is fitted to experimental trajectory statistics to generate long, statistically faithful synthetic trajectories; this model then underpins onboard model predictive control (MPC) that enables the physical device to accurately track a non-trivial target path. The work is presented as a proof of principle for an agentic active particle that senses its state, predicts its evolution, and computes control inputs onboard, in contrast to conventional fixed-motility particles.

Significance. If the quantitative validations hold, the result would be significant for non-equilibrium and soft-matter physics. It demonstrates a concrete integration of physics-informed modeling, data-driven parameter identification, and onboard control that converts an active particle into an adaptive agent. The physical hardware demonstration of MPC trajectory tracking is a clear strength, and the framework is explicitly positioned as a scalable platform for future machine-learning-enabled multi-agent studies in intelligent active matter.

major comments (2)
  1. [Abstract] Abstract: The central claim that the kinematic model 'reproduces measured trajectory statistics' and enables 'accurate execution of a non-trivial target path' is not accompanied by any quantitative error metrics, goodness-of-fit values, or comparison against baselines. This absence is load-bearing because the MPC performance on hardware is asserted to transfer from the digital twin.
  2. [Kinematic model and MPC sections] Kinematic model and MPC sections: The model parameters are identified from passive (uncontrolled) trajectory statistics and then used both to generate synthetic data and to drive the MPC. The manuscript does not report tests of the model's forward prediction accuracy under explicit control inputs (e.g., changes in bristle actuation or motor commands), which is required for reliable closed-loop MPC transfer to hardware.
minor comments (2)
  1. The abstract would be strengthened by including at least one quantitative metric (e.g., RMS path error or diffusion-coefficient match) to support the reproduction and accuracy claims.
  2. Figure captions and text should clarify whether the reported path-tracking results are from a single representative trial or averaged over multiple runs with variability measures.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thoughtful review and for recognizing the potential significance of our work in advancing intelligent active matter. We address the major comments point by point below, providing clarifications and indicating the revisions we will make to the manuscript.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that the kinematic model 'reproduces measured trajectory statistics' and enables 'accurate execution of a non-trivial target path' is not accompanied by any quantitative error metrics, goodness-of-fit values, or comparison against baselines. This absence is load-bearing because the MPC performance on hardware is asserted to transfer from the digital twin.

    Authors: We agree that including quantitative metrics will strengthen the presentation of our results. In the revised manuscript, we will augment the abstract with specific quantitative measures, such as the root-mean-square error for trajectory reproduction and the tracking error for the MPC demonstration. We will also add comparisons to baseline models in the main text to provide context for the goodness of fit. revision: yes

  2. Referee: [Kinematic model and MPC sections] Kinematic model and MPC sections: The model parameters are identified from passive (uncontrolled) trajectory statistics and then used both to generate synthetic data and to drive the MPC. The manuscript does not report tests of the model's forward prediction accuracy under explicit control inputs (e.g., changes in bristle actuation or motor commands), which is required for reliable closed-loop MPC transfer to hardware.

    Authors: The kinematic model is physics-informed and fitted to capture the statistical properties of passive motion, with control inputs entering through modulation of the model parameters based on the device's actuation. While explicit forward prediction tests under controlled inputs were not separately reported, the successful onboard MPC implementation and accurate hardware trajectory tracking provide direct evidence of the model's effectiveness in the closed-loop setting. To further address this point, we will include additional validation of the model's predictive accuracy under varying control inputs in the revised manuscript, using both synthetic and experimental data. revision: yes

Circularity Check

0 steps flagged

No significant circularity; hardware validation provides independent check

full rationale

The paper fits a kinematic model to experimental trajectory statistics to build a digital twin that generates synthetic trajectories, then deploys this model inside onboard MPC whose performance is validated by accurate physical execution of a target path on the device. This hardware demonstration lies outside the fitted statistics and constitutes an external benchmark. No self-definitional equivalence, tautological renaming of fitted outputs as predictions, or load-bearing self-citation chains appear in the described derivation; the central claim of onboard control transfer is not forced by construction from the fitting step alone.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on a kinematic model whose parameters are fitted to trajectory data and on the assumption that this model remains predictive when used inside real-time MPC on the physical hardware.

free parameters (1)
  • kinematic model parameters
    Identified from measured trajectory statistics to match the device's motion; used both for synthetic data generation and for the MPC cost function.
axioms (1)
  • domain assumption The fitted kinematic model produces statistically faithful long trajectories that remain valid for real-time prediction and control on the physical device.
    Invoked when the digital twin is declared the foundation for onboard MPC.

pith-pipeline@v0.9.0 · 5738 in / 1311 out tokens · 39540 ms · 2026-05-22T11:41:31.599438+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

23 extracted references · 23 canonical work pages

  1. [1]

    Bechinger, R

    C. Bechinger, R. Di Leonardo, H. Löwen, C. Reichhardt, G. Volpe, and G. Volpe, Active particles in complex and crowded environments, Rev. Mod. Phys.88, 045006 (2016)

    work page 2016

  2. [2]

    Z. T. Liu, Y. Shi, Y. Zhao, H. Chaté, X. qing Shi, and T. H. Zhang, Activity waves and freestanding vortices in populations of subcritical quincke rollers, Proc. Natl. Acad. Sci. U.S.A.118, e2104724118 (2021)

    work page 2021

  3. [3]

    R. R. Garza, N. Kyriakopoulos, Z. M. Cenev, C. Rigoni, and J. V. I. Timonen, Magnetic quincke rollers with tun- able single-particle dynamics and collective states, Sci. Adv. 9, eadh2522 (2023)

    work page 2023

  4. [4]

    Narayan, S

    V. Narayan, S. Ramaswamy, and N. Menon, Long-lived giant number fluctuations in a swarming granular ne- matic, Science317, 105 (2007)

    work page 2007

  5. [5]

    Scholz, M

    C. Scholz, M. Engel, and T. Pöschel, Rotating robots move collectively and self-organize, Nat. Commun.9, 931 (2018)

    work page 2018

  6. [6]

    Commun.9, 5156 (2018)

    C.Scholz, S.Jahanshahi, A.Ldov,andH.Löwen,Inertial delay of self-propelled particles, Nat. Commun.9, 5156 (2018). 10

    work page 2018

  7. [7]

    Caprini, D

    L. Caprini, D. Breoni, A. Ldov, C. Scholz, and H. Löwen, Dynamical clustering and wetting phenomena in inertial active matter, Commun. Phys.7, 343 (2024)

    work page 2024

  8. [8]

    Chen and J

    Y. Chen and J. Zhang, Anomalous flocking in nonpo- lar granular brownian vibrators, Nat. Commun.15, 6032 (2024)

    work page 2024

  9. [9]

    Caprini, B

    L. Caprini, B. Liebchen, and H. Löwen, Self-reverting vortices in chiral active matter, Commun. Phys.7, 153 (2024)

    work page 2024

  10. [10]

    Dauchot and V

    O. Dauchot and V. Démery, Dynamics of a self-propelled particle in a harmonic trap, Phys. Rev. Lett.122, 068002 (2019)

    work page 2019

  11. [11]

    Baconnier, D

    P. Baconnier, D. Shohat, C. H. López, C. Coulais, V. Démery, G. Düring, and O. Dauchot, Selective and collective actuation in active solids, Nat. Phys.18, 1234 (2022)

    work page 2022

  12. [12]

    O. Chor, A. Sohachi, R. Goerlich, E. Rosen, S. Rahav, and Y. Roichman, Many-body szilárd engine with giant number fluctuations, Phys. Rev. Res.5, 043193 (2023)

    work page 2023

  13. [13]

    Giomi, N

    L. Giomi, N. Hawley-Weld, and L. Mahadevan, Swarm- ing, swirling and stasis in sequestered bristle-bots, Proc. R. Soc. A469, 20120637 (2013)

    work page 2013

  14. [14]

    Zheng, M

    E. Zheng, M. Brandenbourger, L. Robinet, P. Schall, E. Lerner, and C. Coulais, Self-oscillation and synchro- nization transitions in elastoactive structures, Phys. Rev. Lett. 130, 178202 (2023)

    work page 2023

  15. [15]

    Altshuler, O

    A. Altshuler, O. L. Bonomo, N. Gorohovsky, S. Mar- chini, E. Rosen, O. Tal-Friedman, S. Reuveni, and Y. Roichman, Environmental memory facilitates search with home returns, Phys. Rev. Res.6, 023255 (2024)

    work page 2024

  16. [16]

    Noirhomme, I

    M. Noirhomme, I. Mammadli, N. Vanesse, J. Pande, A.- S. Smith, and N. Vandewalle, Brainbots as smart au- tonomous active particles with programmable motion, Phys. Rev. Appl.23, 064008 (2025)

    work page 2025

  17. [17]

    Novkoski, M

    F. Novkoski, M. Mélard, M. Delens, F. Wéry, M. Noirhomme, J. Pande, A. Maier, A. S. Smith, and N. Vandewalle, Graspion: an open-source, pro- grammable brainbot for active matter research (2025), arXiv:2509.07437 [cond-mat.soft]

    work page arXiv 2025

  18. [18]

    Becker, S

    F. Becker, S. Boerner, V. Lysenko, I. Zeidis, and K. Zim- mermann, On the mechanics of bristle-bots - modeling, simulation and experiments, in ISR/Robotik 2014; 41st International Symposium on Robotics (2014) pp. 1–6

    work page 2014

  19. [19]

    Majewski, D

    T. Majewski, D. Szwedowicz, and M. Majewski, Loco- motion of a mini bristle robot with inertial excitation, J. Mech. Robot9, 061008 (2017)

    work page 2017

  20. [20]

    Analytical Chemistry 36(8), 1627–1639 (1964)

    A. Savitzky and M. J. E. Golay, Smoothing and differentiation of data by simplified least squares procedures., Analytical Chemistry 36, 1627 (1964), https://doi.org/10.1021/ac60214a047

    work page doi:10.1021/ac60214a047 1964

  21. [21]

    S. M. Mousavi and V. Fakhari, Design and develop- ment of closed-loop controllers for trajectory tracking of a planar vibration-driven robot, J. Vib. Control31, 3691 (2025)

    work page 2025

  22. [22]

    Nguyen, S

    K. Nguyen, S. Schoedel, A. Alavilli, B. Plancher, and Z. Manchester, Tinympc: Model-predictive control on resource-constrained microcontrollers, in2024 IEEE In- ternational Conference on Robotics and Automation (ICRA) (IEEE, 2024) pp. 1–7

    work page 2024

  23. [23]

    Crank and P

    J. Crank and P. Nicolson, A practical method for numeri- cal evaluation of solutions of partial differential equations of the heat-conduction type, Math. Proc. Camb. Philos. Soc. 43, 50–67 (1947)

    work page 1947