Humanoid Adaptation Framework
Adapting Generalist VLAs to Humanoid Whole-Body Loco-manipulation via Hierarchical Action Flow and Spectral Latent RL
HAF-VLA has been accepted to CoRL 2026 as a separate work.
HAF-VLAHAF-Steer
01 / Overview
Abstract
Humanoid robots hold great promise as general-purpose agents in human-centered environments, yet generalist vision-language-action foundation models are not readily applicable to humanoid whole-body loco-manipulation. High-dimensional and interdependent whole-body motion makes it difficult for conventional single-stage architectures to coordinate locomotion, waist posture, and dual-arm manipulation, while policies trained purely through offline behavior cloning can remain suboptimal during real-world deployment.
HAF, the Humanoid Adaptation Framework, combines two complementary components. HAF-VLA splits full-body action denoising into three sequential stages and preserves kinematic dependencies with stage embeddings and cross-stage KV caches. On top of the frozen generator, HAF-Steer uses flow reversal and DCT compression to train a regularized SAC policy in a compact noise subspace, enabling efficient offline-to-online refinement without updating the large VLA backbone.
02 / Method
Part A
Hierarchical Action Flow
A shared action expert progressively expands the active whole-body action space. Clean actions from earlier stages are encoded as KV caches, conditioning later stages before a single final action chunk is deployed.
Part B
Spectral Latent RL
Expert action chunks are reversed through the frozen flow field, compressed to low-frequency DCT coefficients, and used to initialize a regularized offline-to-online SAC policy.
03 / Real-world tasks
Each task combines navigation, posture adjustment, and physical interaction. All task clips below intentionally play without sound.
Carry + bimanual loading
Navigate + retrieve
Reach + rearrange
Lift + transport
Pick + relocate
Align + throw
Squat + whole-body carry
04 / Experiments
HAF-VLA Performance
| Task | HAF-VLA | pi0.5 | GR00T N1.7 | Cosmos | ACT |
|---|---|---|---|---|---|
| Laundry Loading | 66.7 | 53.3 | 40.0 | 0.0 | 10.0 |
| Clothes Retrieval | 53.3 | 53.3 | 33.3 | 26.7 | 23.3 |
| Table Tidy | 80.0 | 70.0 | 40.0 | 16.7 | 23.3 |
| Basket Transfer | 63.3 | 50.0 | 43.3 | 33.3 | 16.7 |
| Toy Storage | 80.0 | 53.3 | 30.0 | 40.0 | 23.3 |
| Ball Tossing | 56.7 | 33.3 | 36.7 | 3.3 | 30.0 |
| Box Transfer | 93.3 | 60.0 | 43.3 | 73.3 | 50.0 |
| Average | 70.5 | 53.3 | 38.1 | 27.6 | 25.2 |
OOD robustness
With an unseen chair near the Laundry Loading path, HAF-VLA scores 40.0% versus 26.7% for pi0.5. With a 20 cm backward start shift in Clothes Retrieval, it scores 43.3% versus 36.7%.
Offline to online
Across Toy Storage and Basket Transfer with both pi0.5 and HAF-VLA backbones, offline noise behavior cloning provides a strong initialization and offline-to-online SAC delivers further adaptation.
