Translation as a Bridging Action: Transferring Manipulation Skills from Humans to Robots
Summary
This paper proposes a bridging action representation based on relative wrist translation in the head-camera frame to transfer human manipulation skills to bi-manual robots, using a vision-language-action model with interleaved action tokens and attention masking to handle embodiment differences.
View Cached Full Text
Cached at: 06/29/26, 06:01 AM
Paper page - Translation as a Bridging Action: Transferring Manipulation Skills from Humans to Robots
Source: https://huggingface.co/papers/2606.28133
Abstract
Human manipulation skills are transferred to robots more effectively by using a bridging action representation based on relative wrist translation in the initial head-camera frame, combined with a vision-language-action model that handles embodiment differences through interleaved action tokens and attention masking.
We study whether we can learn novel manipulation skills from human actions to a bi-manual robot withparallel grippers. Human action data is cheap, abundant, and diverse, making it one of the most promising resources for scaling up robot learning. Yet transferring skills from humans to robots remains hard: most prior work treats humans as just another bi-manual 6DoF embodiment, where hand-pose estimates are noisy and the contact patterns of human fingers differ fundamentally from those of a parallel gripper. We argue that learning rotation-inclusive action signals from human data is therefore sub-optimal, and instead propose a bridging action representation: therelative wrist translationwithin the initialhead-camera frame, an action space shared by humans and robots. To handle the potential absence of certain action components in different embodiments, we build a π_0-likevision-language-action modelwithinterleaved action tokensandattention masking. On a suite of novelbi-manual manipulationtasks, our bridging action transfers human manipulation knowledge to robots far more effectively than noisy 6DoF human actions and scales with the amount of human data.
View arXiv pageView PDFProject pageAdd to collection
Get this paper in your agent:
hf papers read 2606\.28133
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2606.28133 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2606.28133 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2606.28133 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
BridgeVLA++: A Data-Efficient, Generalizable, and Memory-Augmented Vision-Language-Action Framework for 3D Manipulation
BridgeVLA++ is a memory-augmented vision-language-action framework for 3D robot manipulation that builds on BridgeVLA to add spatio-temporal memory, achieving state-of-the-art results on memory-dependent manipulation benchmarks while preserving data efficiency and generalization.
World-to-Wrist: Task-Conditioned Future Wrist Modeling for Fine-Grained Robot Manipulation
Introduces W2-VLA, a vision-language-action model for fine-grained robot manipulation that models task-conditioned future wrist observations, achieving improved manipulation performance on benchmarks and real-world tasks.
Measuring Language Transfer in Robot Policies: Adding Greek to a Cosmos3 Vision-Language-Action Policy
This study explores adding Greek to a Cosmos3 vision-language-action robot policy, revealing challenges in measurement and showing that bilingual training improves performance but overfits to translator phrasing. The results emphasize the need for null baselines and seed replication in low-resource localization.
iMaC: Translating Actions into Motion and Contact Images for Embodied World Models
iMaC proposes a unified control paradigm that uses raw visual images as native action representations for embodied world models, enabling more expressive and generalized robotic control through image-based action tokens.
See like a Robot: Robot-Centric Pointmaps for Vision-Language-Action Models
This paper introduces robot-centric pointmaps, which encode 3D scene coordinates in the robot frame directly into image pixels, to resolve the frame mismatch between camera observations and robot action definitions in Vision-Language-Action models. The method improves Pi0.5 and SmolVLA on RoboCasa and generalizes better to unseen camera placements in real-robot experiments.