SALA enables dual-axis scaling of embodied foundation models through semantic latent action alignment
Original titleSALA: Semantically Aligned Latent Actions Enable Dual-Axis Scaling of Embodied Foundation Models
- Outlet
- Takara TLDR
- Published
- Importance
- 4/5
Summary
The authors propose SALA to address the challenge of learning latent actions in embodied foundation models. Using language descriptions of unpaired human and robot video clips, SALA estimates semantic similarities and guides human latent actions toward semantically related robot latents. The resulting shared latent action space enables scaling through increasing human video pre-training and task-specific fine-tuning, improving downstream robot task performance.
Original excerpt
Embodied foundation models hinge on diverse robot interaction data, which are costly to collect. Egocentric human videos offer a complementary source of experience, and latent action models (LAMs) provide a popular means of leveraging these videos for policy supervision. Compared with world-action models (WAMs), LAMs can offer greater training efficiency and potentially lower inference costs by avoiding explicit pixel prediction during policy learning and execution. However, most existing LAMs learn latent actions solely through video reconstruction, which can cause their representations to cluster by visual appearance rather than by the underlying actions. As a result, they struggle to align human actions with their robot counterparts. To bridge the gap, we introduce SALA, which emphasizes establishing the semantic correspondence for latent actions. Without requiring paired data, we use the language descriptions of unpaired human and robot clips to estimate semantic similarities and supervise cross-embodiment correspondence. Specifically, we translate these similarities into soft neighborhood structures that guide human latent actions toward semantically related robot latents, while robot action reconstruction grounds the latent space in control. The resulting shared latent action space provides a unified supervision signal for training an embodied foundation model on both human and robot videos. Notably, latent action supervision enables scaling along two axes: (1) pre-training on increasing amounts of general human video consistently reduces robot action prediction error on downstream tasks, and (2) post-training on increasing amounts of task-specific human video substantially improves real-world robot task success rates.