SLA: Scaling Spatial-Language-Action Model via Continual Pre-training
SLA: Scaling Spatial-Language-Action Model via Continual Pre-training
This research presents SLA, a Spatial-Language-Action Model designed to improve reliable robotic manipulation by scaling implicit spatial alignment through continual pre-training.
The paper introduces SLA, a Spatial-Language-Action Model that scales implicit spatial alignment through continual pre-training on heterogeneous robot datasets, utilizing feature-dependent gates and probabilistic state dropout to maintain performance during downstream adaptation while preserving the original inference architecture.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Dataset details reported by the source
- Modalities
- intermediate visual embeddings Open source S3
- Tasks
- robotic manipulation Open source S2
- Embodiments
- diverse embodiments Open source S4
- paper: not_reported
What remains unknown
Dataset licensing and commercial-use rights are not established here.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
from a frozen 3D perception model. We curate and unify heterogeneous robot datasets to support spatial learning across diverse embodiments, tasks, and scenes. To accommodate this diversity, we introduce feature-dependent gates that adaptively aggregate visual embeddings from multiple VLA layers for spatial supervision. The teacher further exploits partially available depth and camera calibration alongside RGB observa
Open source S4
able foundation-model capability therefore calls for learning from broader robot experience. We present SLA, a Spatial-Language-Action Model that scales implicit spatial alignment to continual pre-training and maintains it during downstream post-training. Initialized from a pretrained pi05 policy, SLA jointly optimizes action prediction and alignment between intermediate visual embeddings and spatial representations
Open source S3
arXiv:2611.08373v1 Announce Type: new Abstract: Reliable robotic manipulation requires both semantic understanding and precise control in three-dimensional space. However, vision-language-action (VLA) foundations largely inherit visual representations from 2D vision-language pre-training, while spatial supervision is often introduced only during task-specific adaptation. Developing spatial understanding as a transfer
Open source S2
Implications for data suppliers
RoboSignal interpretation and collection questions, not statements of buyer demand.
- Confirm the required data type and collection setting with the buyer; this source does not establish a complete collection specification.
- Check the exact dataset version, license, modalities and task coverage against the intended use.
- Validate demand and acceptance criteria with a buyer before scaling. Publication, popularity and a research result do not establish a purchase commitment.
Source：arXiv Robotics — research abstracts · arxiv.org