Physics-informed digital twins of brainbots
I. Mammadli, J. Pande ∗, M. Noirhomme, F. Novkoski, A. Maier, N. Vandewalle, A.-S. Smith † · 2025
Read Physics-informed digital twins of brainbots by I. Mammadli, J. Pande ∗, M. Noirhomme, F. Novkoski, A. Maier, N. Vandewalle, A.-S. Smith † (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-00384.txt.
Introduction
Isa Mammadli and eight co-authors built a brainbot: a small programmable bristlebot that senses its state, predicts where it is heading, and computes its own steering on board. Ordinary active particles just move with fixed behavior, so a self-steering particle is a step toward adaptive active matter. Their first step is a physics-informed digital twin of the device.
The twin rests on a kinematic model that reproduces the statistics of measured paths and can generate long, faithful synthetic ones. On that base they run model predictive control directly on the bot, and it follows a non-trivial target path accurately. By fusing physical modeling, data-driven calibration, and control, the platform points toward machine-learning studies of many interacting smart particles.
Catalog record
| CREATOR | I. Mammadli, J. Pande ∗, M. Noirhomme, F. Novkoski, A. Maier, N. Vandewalle, A.-S. Smith † |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 34 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-00384
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Physics-informed digital twins of brainbots I. Mammadli, J. Pande ∗, M. Noirhomme, F. Novkoski, A. Maier, N. Vandewalle, A.-S. Smith † Abstract A brainbot is a robotic device powered by a battery-driven motor that induces horizontal vibrations which lead to controlled two-dimensional motion. While the physical design and capabilities of a brainbot have been discussed in previous work, here we present a detailed theoretical analysis of its motion. We show that the various autonomous trajectories executed by a brainbot – linear, spinning, orbital and helical – are explained by a kinematic model that ascribes angular and translational velocities to the brainbot’s body. This model also uncovers some trajectories that have not so far been observed experimentally. Using this kinematic framework, we present a simulation system that accurately reproduces the experimental trajectories. This can be used to parameterize a digital twin of a brainbot that executes synthetic trajectories that faithfully mimic the required statistical features of the experimental trajectories while being as long as required, such as for machine learning applications. I Introduction Active matter, unlike classical passive systems, consists of energy-consuming units operating far from equilibrium, enabling dynamic behaviors such as self-propulsion, collective motion, and adaptive reorganization. Several paradigmatic artificial designs have appeared over the last decade to facilitate reproducible experimental study. These include systems such as active colloids which are capable of self-propulsion through mechanisms like chemical reactions, light stimuli, and magnetic fields Bechinger et al. 2016 , Quincke rollers that exhibit complex swirling motion Liu et al. 2021 ; Garza et al. 2023 , and vibrated granular rods Narayan et al. 2007 . In addition, there are a variety of active asymmetric particles that move directionally due to imbalanced friction forces resulting from their shape or surface properties. An example of such systems are vibrobots, which are small particles placed on a vertically vibrating bed resulting in their horizontal motion Scholz et al. 2018a ; Scholz et al. 2018b . While they function without internal processing, vibrobots can exhibit interesting self-organization and swarming behaviors Caprini et al. 2024a ; Chen and Zhang 2024 . More recently, chiral vibrobots have been used to demonstrate the role of asymmetric interactions in active self-assembly Caprini et al. 2024b . Another commonly used design of active particles, relying on an internal source of energy for motion generation, are bristlebots, such as the commercially available Hexbugs™ Dauchot and Démery 2019 ; Baconnier et al. 2022 ; Chor et al. 2023 . Bristlebots are equipped with flexible legs and an internal vibration motor powered by a battery, transforming vertical oscillation of the body into horizontal motion Giomi et al. 2013 . This setup allows a considerably larger operational space, and has been used to study active particles in harmonic potentials Dauchot and Démery 2019 , construct active solids Baconnier et al. 2022 and elastoactive structures Zheng et al. 2023 , and mimic ant-like pheromone trail behavior by way of Hexbugs™ moving through fields of passive particles Altshuler et al. 2024 . While bristlebots have seen extensive use, their efficient locomotion has been made programmable and controllable only recently, through the integration of ARM {}^{\text{\textregistered}} microcontrollers into custom-built bristlebots to create brainbots. Brainbots are versatile centimeter-sized devices that include light, color and magnetic field sensors and infrared emitter-receivers Noirhomme et al. 2025 ; Novkoski et al. 2025 . Powered by a lithium-ion battery allowing 40 - 60 minutes of operation, these robots exhibit cycloidal motion patterns, which can be programmed to exhibit ballistic and diffusive motions. Their trajectories have been found to be strongly sensitive to the inclination of their legs and the motor power, and various kinds of autonomous trajectories have been observed, such as linear, spinning, orbital and helical Noirhomme et al. 2025 . Understanding the motion of bristlebots and brainbots is necessary to improve the control of their trajectories and enhance their performance and applicability. Physical models of bristlebot locomotion, based on the flexibility of the legs, have been introduced in one dimension Giomi et al. 2013 ; Becker et al. 2014 and two dimensions Majewski et al. 2017 . However, these have not been successfully combined with experimental data, nor do they account for the diversity of trajectories that such robots can exhibit Noirhomme et al. 2025 . In this paper we propose a framework to describe the various trajectories executed by a brainbot, which can be easily extended to more complex settings and applied to control algorithms. While in our previous work the different observed trajectories were distinguished using a parameter \eta which quantified how similar an observed motion was to spinning Noirhomme et al. 2025 , a theoretical framework to capture the relevant kinematics was missing. In this paper we provide this kinematic framework to explain and decouple the translational, spinning and orbital modes of brainbot trajectories observed in the experiments. We then implement our theoretical framework in simulations, and use an optimization procedure to find the parameters that reproduce the complicated experimental trajectories to a high degree of accuracy. We then utilize the statistical distributions and Fourier modes of the trajectory parameters in order to generate long artificial trajectories that have similar statistical properties to the experimental ones, thus creating a digital twin of the brainbot. II Kinematics of brainbot motion The brainbot moves predominantly in two dimensions, here taken to be the x - y plane. The brainbot surface is an ellipse, whose geometric center is described by the two-dimensional position vector \bm{r}(t) in the laboratory frame at time t . The orientation of the brainbot is described by the angle \varphi(t) that the major axis of the brainbot makes with the horizontal x -axis at time t (Fig. 1 (a)). Figure 1: (a) Position vector \bm{r}(t) and orientation angle \varphi(t) of a brainbot at time t , in the global reference frame. (b) Position vectors of the geometric center \bm{r} of the brainbot and an arbitrary point \bm{r}^{\prime} in the laboratory frame, and the instantaneous unit vectors \bm{\hat{n}_{1}} and \bm{\hat{n}_{2}} along the major and minor axis, respectively. (c) Semi-major and semi-minor axis lengths A_{1} and A_{2} , and the coordinates of the geometric center (0,0) and an arbitrary point ( \rho_{1}A_{1},\rho_{2}A_{2} ) in the body-fixed reference frame. Let A_{1} and A_{2} be the semi-major and semi-minor axis lengths of the brainbot ellipse, and let \bm{\hat{n}_{1}} and \bm{\hat{n}_{2}} be the unit vectors, in the laboratory frame, along the major and minor axes of the ellipse, given by \displaystyle\bm{\hat{n}_{1}}(\varphi) \displaystyle=(\cos\varphi,\ \sin\varphi), (1) \displaystyle\bm{\hat{n}_{2}}(\varphi) \displaystyle=(-\sin\varphi,\ \cos\varphi). (2) Let \bm{r}^{\prime} denote an arbitrary point in the plane (inside or outside the ellipse) in the laboratory frame, and let its coordinates be \rho_{1}A_{1} and \rho_{2}A_{2} in the reference frame of the brainbot, where \rho_{1} and \rho_{2} are scale factors. If \bm{r}^{\prime} is on or inside the ellipse, then the \rho ’s must lie between -1 and 1 , and must in fact satisfy the condition \rho_{1}^{2}+\rho_{2}^{2}\leq 1. (3) The coordinates of the geometric center \bm{r} and the arbitrary point \bm{r}^{\prime} are related by \bm{r}^{\prime}=\bm{r}+\rho_{1}A_{1}\bm{\hat{n}_{1}}+\rho_{2}A_{2}\bm{\hat{n}_{2}}, (4) while their velocities ( \bm{v}^{\prime} at \bm{r}^{\prime} and \bm{v} at \bm{r} ) are related by \bm{v}^{\prime}=\bm{v}+\bm{\omega}\times(\bm{r}^{\prime}-\bm{r}), (5) where \bm{\omega}\equiv(0,0,\omega) is the angular velocity vector. Let {\bm{r}_{\rm c}}(t) denote the instantaneous center of rotation of the brainbot,