H-Zero: Cross-Humanoid Locomotion Pretraining Enables Few-shot Novel Embodiment Transfer

2025/11/30 by Yunfeng Lin, Minghuan Liu, Lin, Yunfeng +11
Engineering · Psychology · #FOS: Computer and information sciences #Human Motion and Animation #Robotic Locomotion and Control #Robotics (cs.RO) #Social Robot Interaction and HRI

paper · pdf · doi:10.48550/arxiv.2512.00971

openalex publication_date 2025/11/30 · openalex created_date 2025/12/03 · openalex updated_date 2026/07/28

Abstract

The rapid advancement of humanoid robotics has intensified the need for robust and adaptable controllers to enable stable and efficient locomotion across diverse platforms. However, developing such controllers remains a significant challenge because existing solutions are tailored to specific robot designs, requiring extensive tuning of reward functions, physical parameters, and training hyperparameters for each embodiment. To address this challenge, we introduce H-Zero, a cross-humanoid locomotion pretraining pipeline that learns a generalizable humanoid base policy. We show that pretraining on a limited set of embodiments enables zero-shot and few-shot transfer to novel humanoid robots with minimal fine-tuning. Evaluations show that the pretrained policy maintains up to 81% of the full episode duration on unseen robots in simulation while enabling few-shot transfer to unseen humanoids and upright quadrupeds within 30 minutes of fine-tuning.

Citations

• UniTracker: Learning Universal Whole-Body Motion Tracker for Humanoid Robots
• A Unified and General Humanoid Whole-Body Controller for Versatile Locomotion
• Multi-Loco: Unifying Multi-Embodiment Legged Locomotion via Reinforcement Learning Augmented Diffusion
• Towards Embodiment Scaling Laws in Robot Locomotion
• BeamDojo: Learning Agile Humanoid Locomotion on Sparse Footholds
• Learning Humanoid Standing-up Control across Diverse Postures
• ExBody2: Advanced Expressive Humanoid Whole-Body Control
• HOVER: Versatile Neural Whole-Body Controller for Humanoid Robots
• One Policy to Run Them All: an End-to-end Learning Approach to Multi-Embodiment Locomotion
• Scaling Cross-Embodied Learning: One Policy for Manipulation, Navigation, Locomotion and Aviation
• GET-Zero: Graph Embodiment Transformer for Zero-shot Embodiment Generalization
• Advancing Humanoid Locomotion: Mastering Challenging Terrains with Denoising World Model Learning
• OmniH2O: Universal and Dexterous Human-to-Humanoid Whole-Body Teleoperation and Learning
• Learning Human-to-Humanoid Real-Time Whole-Body Teleoperation
• Expressive Whole-Body Control for Humanoid Robots
• Agile But Safe: Learning Collision-Free High-Speed Legged Locomotion
• Extreme Parkour with Legged Robots
• ANYmal Parkour: Learning Agile Navigation for Quadrupedal Robots
• Barkour: Benchmarking Animal-level Agility with Quadruped Robots
• DreamWaQ: Learning Robust Quadrupedal Locomotion With Implicit Terrain Imagination via Deep Reinforcement Learning
• Multi-embodiment Legged Robot Control as a Sequence Modeling Problem
• Sim-to-Real Transfer for Quadrupedal Locomotion via Terrain Transformer
• Walk These Ways: Tuning Robot Control for Generalization with Multiplicity of Behavior
• Deep Whole-Body Control: Learning a Unified Policy for Manipulation and Locomotion
• Learning and Deploying Robust Locomotion Policies with Minimal Dynamics Randomization
• GenLoco: Generalized Locomotion Controllers for Quadrupedal Robots
• A Walk in the Park: Learning to Walk in 20 Minutes With Model-Free Reinforcement Learning
• DayDreamer: World Models for Physical Robot Learning
• Minimizing Energy Consumption Leads to the Emergence of Gaits in Legged Robots
• Isaac Gym: High Performance GPU-Based Physics Simulation For Robot Learning
• RMA: Rapid Motor Adaptation for Legged Robots
• Learning Modular Robot Control Policies
• Dynamics Randomization Revisited:A Case Study for Quadrupedal Locomotion
• Crossing The Gap: A Deep Dive into Zero-Shot Sim-to-Real Transfer for Dynamics
• One Policy to Control Them All: Shared Modular Policies for Agent-Agnostic Control
• Learning Agile Robotic Locomotion Skills by Imitating Animals
• Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learning
• Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World
• Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)

Related

• BFM-Zero: A Promptable Behavioral Foundation Model for Humanoid Control Using Unsupervised Reinforcement Learning
• Human-Humanoid Robots Cross-Embodiment Behavior-Skill Transfer Using Decomposed Adversarial Learning from Demonstration
• BeyondMimic: From Motion Tracking to Versatile Humanoid Control via Guided Diffusion
• DemoHLM: From One Demonstration to Generalizable Humanoid Loco-Manipulation
• Robot Trains Robot: Automatic Real-World Policy Adaptation and Learning for Humanoids