A three-stage pipeline that turns motion-imitation skills into a reusable hybrid motion prior (HMP) for humanoid locomotion and shows that training the codebook with the rotation trick improves latent organization and reduces downstream falls compared with a standard straight-through estimator.
Abstract
Reinforcement learning can produce robust humanoid controllers, but each new task is typically trained as a separate policy with its own reward design and training process. Motion imitation provides an alternative source of motor competence by training policies to track retargeted human motions, yet the resulting controllers remain reference trackers and are not directly usable as task policies. We propose a three-stage pipeline that turns motion-imitation skills into a reusable hybrid motion prior (HMP) for humanoid locomotion. First, an expert policy is trained to imitate retargeted human motion-capture clips. Second, the expert is distilled into a frozen architecture composed of a proprioceptive encoder, a residual vector-quantized (RVQ) codebook, and an action decoder. Third, task-level policies are trained to solve locomotion tasks by selecting discrete codebook entries while the HMP remains frozen. We evaluate the method on velocity tracking, point-goal navigation, and fall-recovery velocity tracking in simulation, and deploy the velocity-tracking policy on a real Unitree G1 robot. The distillation process preserves the tracking behavior of the expert, while the resulting HMP can be reused without retraining as the action interface for different downstream locomotion policies. The learned HMP reveals an interpretable codebook structure in which the number of active RVQ stages modulates the available gait patterns. We further show that training the codebook with the rotation trick improves latent organization and reduces downstream falls compared with a standard straight-through estimator.
Humanoid general motion tracking requires policies that can follow diverse whole-body references while maintaining balance. Building such policies involves many practical design choices, and their individual effects are often hard to assess. We address this issue with an empirical study of common modeling and training factors used in recent humanoid motion-imitation pipelines. To make the study controlled and reproducible, we developed YAHMP, an open-source modular framework for training, evaluating, and deploying whole-body motion tracking policies on the Unitree G1. Within YAHMP, we define a nominal configuration and compare variants that differ in motion-command representation, observation history, action representation, actuation profile, hand-force randomization during training, and training approach. We evaluate the resulting policies on a test set of retargeted human motions and compare the nominal policy with TWIST2 as an external baseline trained on the same motion set. The results distinguish choices with clear tracking effects from choices that mainly change actuation effort, training complexity, or physical interaction capability. Finally, we deploy YAHMP policies zero-shot on the real Unitree G1, demonstrating diverse whole-body motion tracking, balance under external perturbations, and forceful interaction.
Humanoid robots have recently demonstrated promising capabilities in real-world ball sports. However, achieving professional motion styles while maintaining strong task performance remains challenging. In this work, we propose AdaPT, an Adaptive Motion Planning and Tracking framework that learns professional tennis serving and rally styles directly from broadcast videos. This hierarchical design is motivated by the key insight that the planner generates stylistic kinematic motions, while the tracker executes them with minimal interference with planning. Despite its effectiveness in simulation, a substantial sim-to-real gap emerges: tracking performance inevitably degrades on real robots, and this degradation is partially overlooked by autoregressive planning and further compounded by noisy perception. To address these issues, our adaptation mechanism improves tracking robustness by learning to track randomized execution speeds, while conditioning the planner on a learned motion-speed adapter to mitigate compounding errors. Real-world experiments on the Unitree G1 demonstrate the effectiveness of our adaptation mechanism in bridging the sim-to-real gap. We further deploy AdaPT policies on the full-size Dobot Atom humanoid robot (1.7m) and demonstrate in-the-wild serving without motion capture. Beyond these results, our real-world experiments reveal both algorithmic and engineering insights for future humanoid ball-sports systems. Videos and code are available on our \href{https://humanoidtennis.github.io/AdaPT/}{project website}.
Tao Huang, Ruofei Liu, Xuchen Tang et al.· 0 citations
Recent work in humanoid whole-body control has found success with a simple recipe: retarget human motion to robot kinematic references, then train policies via reinforcement learning (RL) to track them. But how does this recipe transfer to dexterous manipulation? The answer is not obvious, as manipulation involves complex, contact-rich dynamics and requires delicate regulation of contact modes and forces. We present REGRIND, a minimalist retargeting-guided RL pipeline that learns dexterous manipulation policies from a single human demonstration. REGRIND retargets human hand-object motion to a robot reference that preserves hand-object spatial and contact relationships, trains a residual RL policy in simulation to track object-centric keypoints along that reference, and transfers the resulting policy zero-shot to hardware with careful system identification. The resulting policies produce fluid, human-like behavior on two different multi-fingered hands across contact-rich tool-use tasks, including operating a pair of scissors and turning a screwdriver. Through systematic hardware experiments, we identify and analyze the key factors that govern sim-to-real transfer in dexterous manipulation, offering practical guidance for retargeting-based learning in contact-rich settings. Videos and code are available at https://yunhaifeng.com/REGRIND.
Yunhai Feng, Natalie Leung, Jiaxuan Wang et al.· 0 citations
This work categorizes quadruped robot skills into three types: rhythmic motions, expressive motions, and high-dynamic motions and generates reference trajectories for each category using central pattern generators, animation design, and motion capture, respectively, and designs an asymmetric neural network architecture and employs an imitation–reinforcement learning algorithm to train policies for generating these three types of motions.
Chong Pi, Senwei Huang, Wei Li et al.· Robotica (Cambridge. Print)· 0 citations
Robot skill generation is often approached from two distinct perspectives: normative trajectory optimization, which emphasizes smoothness-based criteria such as minimum jerk, and imitation-based learning, which prioritizes fidelity to demonstrated behaviors. While both paradigms aim to produce feasible and meaningful motions, they are typically formulated separately. In practice, however, many robotic skills require trajectories that are both dynamically smooth and faithful to demonstrations. We propose a unified framework for normative–imitative trajectory optimization that makes this trade-off explicit and tunable. Our framework formulates trajectory generation as a constrained quadratic program combining weighted linear-operator smoothness penalties, a quadratic imitation anchoring term, and affine equality constraints for feasibility. For affine-constrained instances, the resulting problem is strictly convex, admits a unique global minimizer, and can be solved efficiently using standard quadratic programming techniques. Our proposed formulation unifies a broad class of smoothness objectives, including minimum velocity, acceleration, jerk, snap, and elastic energy models, within a single operator-based representation, while incorporating demonstration fidelity in a principled manner. Simulation and real-world experiments on a UR5e robotic arm demonstrate that our framework provides predictable interpolation between purely normative and purely imitative behaviors, offering a compact and extensible foundation for trajectory learning from demonstration.
Reza Azadeh· 2026 23rd International Conf...· 0 citations
A framework for learning human-like robot motion from demonstration, including data collection, probabilistic trajectory learning, and perceptual user evaluation is presented, extending the widely used Gaussian Mixture Model and Gaussian Mixture Regression approach for learning from demonstration.
Alperen Kenan, Paul A. Bremner, Manuel Giuliani· 1 citation