asRoBallet: Closing the Sim2Real Gap via Friction-Aware Reinforcement Learning for Underactuated Spherical Dynamics

2026-04-27

AI Summary

This work addresses the Sim2Real challenges—such as complex friction dynamics, actuator latency, and constrained safe exploration—that hinder the deployment of reinforcement learning (RL) on physical humanoid spherical robots. To bridge this gap, the authors develop a high-fidelity MuJoCo simulation environment that explicitly models the contact mechanics of omnidirectional wheels with discrete rollers. They further introduce a friction-aware RL framework that accurately captures the multi-channel frictional coupling among the wheel, sphere, and ground. Combined with a subtraction-based mechanical redesign and a low-latency human–robot interface built from consumer-grade electronics, this approach enables, for the first time, zero-shot Sim2Real transfer of RL policies for such robots. The system allows operators to intuitively command complex humanoid maneuvers through natural body motions, with real-world experiments validating its stability and expressive capabilities.

Abstract

We introduce asRoBallet, to the best of our knowledge, the first successful deployment of reinforcement learning (RL) on a humanoid ballbot hardware. Historically, ballbots have served as a canonical benchmark for underactuated and nonholonomic control, which are characterized by a reality gap in complex friction models for wheel-sphere-ground interactions. While current literature demonstrates successful handling of 3D balancing with LQR and MPC, 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, and safe hardware exploration. This study proposes a high-fidelity MuJoCo simulation that explicitly models the discrete roller mechanics of ETH-type omni-wheels, thereby capturing parasitic vibrations and contact discontinuities that are previously ignored. We also developed a Friction-Aware Reinforcement Learning framework that achieves zero-shot Sim2Real transfer by mastering the coupled rolling, lateral, and torsional friction channels at the wheel-sphere and sphere-ground interfaces. We designed asRoBallet through subtractive reconfiguration, repurposing key components from an overconstrained quadruped and integrating them into a newly designed structural frame to achieve a robust research platform at low cost. We also developed a generalized iOS ecosystem that transforms consumer electronics into a low-latency interface, enabling a single operator to orchestrate expressive humanoid maneuvers via intuitive natural motion.

Problem

• Sim2Real gap
• underactuated spherical dynamics
• friction modeling
• humanoid ballbot
• reinforcement learning

Innovation

• Friction-Aware Reinforcement Learning
• Sim2Real Transfer
• Underactuated Spherical Dynamics
• High-Fidelity Simulation
• Ballbot Control