---
title: 'asRoBallet: RL-Controlled Humanoid Ballbot'
url: https://www.emergentmind.com/topics/asroballet
type: topic
---

# asRoBallet: RL-Controlled Humanoid Ballbot

Searching arXiv for the asRoBallet paper and closely related robotic choreography/control work.
asRoBallet is a humanoid ballbot system introduced as “the first successful deployment of reinforcement learning (RL) on a humanoid ballbot hardware,” with the stated objective of closing the Sim2Real gap for underactuated and nonholonomic control in ballbots through friction-aware modeling, domain-randomized RL, and a low-cost hardware platform derived from an overconstrained quadruped [2604.24916]. The system integrates a high-fidelity MuJoCo model of wheel–sphere–ground contact, a PPO-based control framework for velocity tracking and station keeping, a subtractive-reconfigured hardware embodiment with 49 DoF, and a generalized iOS interface for single-operator orchestration of expressive humanoid maneuvers [2604.24916]. In the broader literature on robotic choreography and expressive robotic systems, asRoBallet sits at the intersection of underactuated mobile robotics, sim-to-real transfer, and movement-centered human-robot interaction [1712.08195], while differing from prior robot dance systems that focused on industrial arms [2411.12361], humanoid dance generation [1707.01489], or dance-training interaction [1810.12462].

## 1. System identity and research problem

asRoBallet addresses a long-standing control problem associated with ballbots, which the source describes as “a canonical benchmark for underactuated and nonholonomic control,” and identifies as being limited by “a reality gap in complex friction models for wheel-sphere-ground interactions” [2604.24916]. The paper states that current literature had already demonstrated “successful handling of 3D balancing with LQR and MPC,” but that “transitioning to actual hardware for a humanoid ballbot using RL is currently hindered by critical gaps in contact modeling, actuator latency & jitter, and safe hardware exploration” [2604.24916].

The system is presented as a tightly coupled combination of three components: “(1) a very high-fidelity MuJoCo simulation of underactuated spherical dynamics with explicit discrete roller mechanics, (2) a friction-aware reinforcement-learning framework that masters coupled rolling, lateral and torsional friction channels, and (3) a hardware embodiment built by subtractive reconfiguration of a quasi-direct-drive quadruped and controlled via a consumer-grade iOS ecosystem” [2604.24916]. This framing is central to the asRoBallet contribution: the work is not only an RL controller for balancing, but an end-to-end platform spanning simulator fidelity, learning design, hardware embodiment, and teleoperation interface.

A plausible implication is that the name “asRoBallet” serves a double role. It identifies the specific humanoid ballbot platform of [2604.24916], while also resonating with a wider body of robot performance and choreographic research in which movement expressivity, audience interpretation, and robot embodiment are treated as technical design variables rather than merely artistic superstructure [1712.08195], [2411.12361].

## 2. High-fidelity MuJoCo model and frictional contact structure

The simulation model is designed to reproduce contact phenomena that simpler ballbot abstractions omit. In MuJoCo, asRoBallet is modeled “as a floating-base multibody comprising the torso, three ETH-type omni-wheel drive units, and a hollow steel sphere coated with SBR rubber” [2604.24916]. Each omni-wheel is represented as “a rigid ring fitted with 12 free-spinning rollers, alternating between load-bearing and spacer rollers, connected by hinge joints” [2604.24916]. According to the source, this “discrete geometric fidelity is essential to reproduce lateral compliance, parasitic vibrations during roller handovers, and contact discontinuities” [2604.24916].

At the sphere-ground interface, the model enforces “full six-dimensional contact: a normal force $N$, two tangential sliding directions, torsional spin, and two rolling resistance channels” [2604.24916]. The rolling resistance is specified as
$$
F_{\mathrm{roll} = \mu_{\mathrm{roll}\,N
$$
with the additional damping term
$$
M_{\mathrm{roll},i} = -c_r\,\dot\theta_{r,i}\,,\;i\in\{x,y\}
$$
which is used “to capture parasitic vibrations during roller handover” [2604.24916]. Tangential directions use Coulomb friction,
$$
F_{t} = \mu_{\mathrm{slide}\,N\,\mathrm{sign}(v_{t})\,,
$$
and torsional friction is modeled as
$$
M_{\mathrm{torsion} = \begin{cases}
\mu_{\mathrm{torsion}\,N &\text{if }\omega_{z}\neq0,\\
\le \mu_{\mathrm{torsion}\,N &\text{if }\omega_{z}=0
\end{cases}
$$
to represent stick-slip discontinuity [2604.24916].

The ball–roller interface uses “standard three-dimensional contact with a soft penetration law”
$$
\mathrm{solimp}=(0.85,0.99,0.003)
$$
so that “the rubber layer stiffens over 3 mm, reproducing hysteresis and anisotropic slip arising purely from roller compliance” [2604.24916]. Motor-side actuator friction is identified experimentally through
$$
I_\mathrm{m}\,\ddot\theta = \tau - D_v\,\dot\theta - D_c\,\mathrm{sign}(\dot\theta),
$$
where “$D_c\approx130$ mNm is the Coulomb breakaway torque measured experimentally” and the resulting dead-zone torque “creates micro-oscillatory limit cycles in station keeping” [2604.24916].

The simulator configuration is also specified in detail: MuJoCo uses the “elliptic friction cone (cone=‘elliptic’), a stiff constraint impedance ratio (impratio=10), and the Newton solver with noslip_iterations=1 to faithfully resolve stick-slip transitions at 0.2 ms internal steps (frame_skip=5 for a 0.01 s control period)” [2604.24916]. The paper’s central claim is that explicit tribological modeling of rolling, sliding, torsional friction, and roller handover dynamics is necessary for zero-shot transfer on this class of underactuated platform [2604.24916].

## 3. Friction-aware reinforcement learning framework

The RL framework trains “two base policies—velocity tracking and station keeping—over this high-fidelity model using a standard on-policy actor-critic algorithm (PPO)” [2604.24916]. The action space is
$$
a_t\in\mathbb R^3
$$
and consists of “direct torque commands $\tau$ to the three omni-wheels” [2604.24916].

For velocity tracking, the observation is
$$
o_t^{\mathrm{VT}=[\,v_B^{x,y},\;\hat v^{x,y},\;q_{\mathrm{tilt},\;\omega_B^{x,y,z},\;a_{t-1}\,]^\top
$$
and includes “body-frame linear velocity $v_B$, commanded velocity $\hat v$, tilt quaternion $q_{\rm tilt}$, gyro rates $\omega_B$, and previous action” [2604.24916]. For station keeping, the observation appends “world-frame position error $p_W\in\mathbb R^2$ and heading error $\psi$” [2604.24916].

The reward combines exponential tracking terms,
$$
r_v=\exp\bigl(-\|v_B-\hat v\|^2/\sigma_v^2\bigr),\quad
r_\omega=\exp\bigl(-(\omega_z-\hat\omega_z)^2/\sigma_\omega^2\bigr),
$$
penalties
$$
-\lambda_{\rm ang}\|\omega_{xy}\|^2,\quad
-\lambda_\tau\|\tau\|^2,\quad
-\lambda_{\rm smooth}\|a_t-a_{t-1}\|^2,
$$
and an “uprightness constraint $\mathbb 1(|\phi|,|\theta|\le20^\circ)$” [2604.24916]. For station keeping, the velocity-tracking terms are replaced with “position and heading terms $1-\tanh(\|[x,y,\psi]\|)$” [2604.24916]. The total return is
$$
J(\theta)=\mathbb E_{\tau\sim\pi_\theta}\Bigl[\sum_{t=0}^T\gamma^t\,r_t\Bigr]
$$
and is optimized “with standard PPO clipping and an $\ell_2$ weight penalty” [2604.24916].

A distinctive feature of the method is friction-aware domain randomization. The paper states that “friction parameters $\mu_{\mathrm{slide},\mu_{\mathrm{torsion},D_c$ and IMU noise $\mathcal N(0,0.009^2)$ are domain-randomized each episode” according to
$$
\mu\sim\mathcal U(0.6,1.2),\;D_c\sim\mathcal U(0.08,0.12)\,\rm Nm.
$$
[2604.24916] The source further notes that “a virtual low-damped ball joint between body and sphere enforces contact throughout training; ablations show that without this coupling the reward landscape is too sparse and training fails to converge” [2604.24916].

This design differentiates asRoBallet from earlier robot dance generation systems that focused on latent-variable movement synthesis from music [1707.01489] or choreographic motion authoring [1712.08195]. Here, the central learning problem is robust control under highly coupled contact uncertainty rather than sequence generation for stylistic novelty.

## 4. Hardware embodiment and iOS control ecosystem

The hardware platform is produced by “subtractive reconfiguration” of the open-source “asOverDog quadruped” [2604.24916]. The “overconstrained four-leg drivetrains become three 5.5 Nm RobStride05 actuators driving ETH omni-wheels around the sphere,” while “the higher-torque 17 Nm RobStride01 actuators reassemble into a 2-DoF head and two 4-DoF arms” [2604.24916]. The robot retains “the original NUC + Jetson Orin computational core, a 2 kHz IMU and current-sensing quasi-direct-drive actuators for intrinsic proprioception” [2604.24916].

The structural design is organized to reduce inertia about the center of mass: “A T-shaped sheet-metal shoulder frame and 3D-printed connectors minimize inertia about the CoM” [2604.24916]. The paper reports that the “final robot has 49 DoF (13 active, 36 passive) and costs an order of magnitude less than custom ballbot drivetrains” [2604.24916]. The passive DoF reflect the compliant and rolling contact elements of the system rather than purely articulated anthropomorphic joints.

The whole-body interface uses commodity mobile hardware. An “iPhone 12 Pro mounted on the spine runs ARKit VIO at 100 Hz, delivering drift-free 6 DoF pose with sub-5 % relative position error and ~1.2 ms peer-to-peer latency via Wi-Fi Aware” [2604.24916]. A “custom iOS app unifies three control modalities—from touchscreen joystick to wearable IMU suit to markerless whole-body pose tracking across iPhone, Watch and AirPods—allowing a single operator to teleoperate asRoBallet’s base and upper body in intuitive, expressive manners” [2604.24916].

This hardware-software integration places asRoBallet within a broader trend in robot performance systems that couple control architecture to choreographic usability. Comparable concerns appear in industrial-arm performance systems using teach mode and force-mode improvisation [2411.12361], and in movement-centered interface design for expressive robotic systems [1712.08195]. In asRoBallet, however, the control interface is directly linked to an underactuated mobile base whose primary stabilization policy is learned rather than scripted [2604.24916].

## 5. Experimental results and zero-shot Sim2Real transfer

The simulation results are reported under “four increasing levels of difficulty (random friction, random tilt, random arm payload)” [2604.24916]. Under these conditions, “the RL policy maintains 100 % velocity-tracking success with MAE 4.0 cm/s (STD 7.1 cm/s) and perfect station keeping (4.5 cm range, 3.0 cm/s residual), whereas a tuned LQR baseline falls to 56 % success with MAE>23 cm/s under random orientation and fails station keeping entirely” [2604.24916].

The ablation results identify two ingredients as essential: “both friction randomization and IMU noise injection are essential: removing either drops tracking success to ~80 %” [2604.24916]. This finding is consistent with the paper’s central thesis that the reality gap is dominated not by gross rigid-body mismatch but by frictional uncertainty and sensing imperfections.

On hardware, the paper reports “zero-shot Sim2Real across ceramic, carpet, slope and bump transitions” [2604.24916]. For station keeping, the robot “drifts less than 5 cm on tile and 3 cm on a yoga mat, with residual speeds ~5 cm/s” [2604.24916]. In “push-recovery trials (0.3 m lateral pushes),” the system “succeed[s] 7/7 times” [2604.24916], and “velocity tracking under human joystick control yields MAE 0.05 m/s” [2604.24916]. The robot is also reported to “survive repeated falls and retain serviceability within seconds” [2604.24916].

These results sharply distinguish asRoBallet from prior dance-oriented robotic systems whose technical emphasis was not sim-to-real stabilization. For example, “Creative Robot Dance with Variational Encoder” focuses on real-time generation of “new dancing movements according to to the listened music” on a NAO humanoid [1707.01489], while “Breathless” emphasizes sinusoidal motifs, pose extraction with OpenPose, and live teach-mode interaction on a UR5e industrial arm [2411.12361]. asRoBallet instead uses expressive motion as a downstream capability built on robust underactuated locomotor control [2604.24916].

## 6. Position within robotic choreography and expressive robotics

asRoBallet belongs to a broader research landscape in which movement is treated as a computational, perceptual, and embodied medium. “Choreographic and Somatic Approaches for the Development of Expressive Robotic Systems” argues that “professional choreographers, dance practitioners, and movement analysts are critical to research in robotics,” and introduces frameworks based on Body, Effort, Space, and Shape, together with embodied exercises and rapid-prototyping tools for movement-centered design [1712.08195]. That line of work is concerned with how movement qualities are interpreted by human viewers and how movement vocabularies can be formalized for robot control [1712.08195].

Other related work explores different technical slices of robotic dance and performance. “Creative Robot Dance with Variational Encoder” uses a VAE with “latent dimension $L=2$” driven by music features such as “loudness” and “variance” to generate upper-body dance motion for a NAO humanoid [1707.01489]. “Dance Teaching by a Robot” combines “adaptive impedance-based controller” design with “Progressive Teaching (PT)” scoring for partner-dance training, showing statistically significant gains in “comfort,” “peace of mind,” and “robot performance” at the $p<.01$ level [1810.12462]. “Breathless” presents an eight-hour duet between a human dancer and a UR5e using per-joint sinusoidal trajectories, OpenPose-based motion extraction, teach mode, and force-based cueing [2411.12361].

Within this landscape, asRoBallet contributes a different axis of progress: not movement symbolism, audience interpretation, or pedagogical haptics, but the resolution of friction-sensitive underactuated dynamics sufficiently well that RL can be deployed “zero-shot” on a humanoid ballbot hardware [2604.24916]. This suggests a possible convergence between expressive robotics and contact-rich mobile control: once stable and transferable balance is available, choreographic frameworks from expressive robotics [1712.08195] and movement-generation systems [1707.01489] could, in principle, be layered atop a ballbot substrate.

A common misconception would be to treat asRoBallet primarily as a robot-performance project in the same sense as robot dance installations or theatrical systems. The paper’s actual emphasis is narrower and more technical: it is a Sim2Real control study centered on friction modeling, PPO training, and underactuated spherical dynamics, with expressive teleoperation and humanoid embodiment included as consequential platform features rather than the sole scientific endpoint [2604.24916].

## 7. Significance, limitations, and research implications

The principal significance claimed for asRoBallet is that it “closes the long-standing reality gap for underactuated spherical dynamics” by “explicitly simulating discrete roller mechanics, modeling coupled friction channels, and weaving domain-randomized RL with a consumer-grade iOS control loop” [2604.24916]. The paper further states that the “subtractive-reconfigured hardware and all-in-one iOS interface democratize humanoid ballbots for future research in dynamic mobile manipulation and multi-agent coordination” [2604.24916].

Several limitations are implicit in the reported scope. The RL framework is described in terms of “two base policies—velocity tracking and station keeping” [2604.24916], rather than a unified whole-body manipulation policy. The strongest empirical claims concern balancing, velocity tracking, station keeping, and push recovery across surface transitions [2604.24916]. This suggests that the work establishes a control foundation rather than a complete autonomous choreography system. A plausible implication is that future extensions would need to integrate the robust base policies with higher-level motion organization, embodied movement vocabularies, or interaction objectives of the kinds studied in expressive robotics [1712.08195], dance generation [1707.01489], or performance systems [2411.12361].

In that sense, asRoBallet can be understood as a control-centric foundation for expressive humanoid ballbots. Its novelty lies in treating wheel–sphere–ground tribology, actuator friction, observation noise, and embodied hardware redesign as inseparable components of a single Sim2Real pipeline [2604.24916]. For researchers interested in underactuated mobile manipulation, physically expressive robotics, or low-cost high-agility platforms, the work is significant not because it replaces prior choreographic or artistic systems, but because it makes a previously fragile embodiment class available to RL-based deployment on real hardware [2604.24916].

Source: https://www.emergentmind.com/topics/asroballet