# Robota-Embed boosts robot task success with unified vision‑language‑action embeddings

> The new model retrieves past robot experiences to improve in‑context learning, raising success rates on simulation and real‑world tasks.

Oossa · 2026-10-15 · https://oossa.com/en/robota-embed-boosts-robot-task-success-with-unified-vision-language-action-embed

Researchers led by Yunze Liu released Robota-Embed, a vision‑language‑action embedding model for robots. It encodes current state, visual input and actions together, then retrieves matching past continuations as context. In tests, it lifted success rates from 47% to 56% on the DROID‑Sim benchmark and from 86% to 93% on the LIBERO difficult subset. The model also helped a frozen GPT‑6 Astra policy succeed on five ManiSkill tasks at 68% versus 24% for the policy alone.

## The facts

- Training corpus: 1,695,684 cleaned anchors; benchmark: 4,000 queries, 8,000‑entry gallery
- Published: 2026-10-15

## Why it matters

Robot developers can use Robota-Embed to improve task performance without retraining policies, saving compute and data collection effort.

## Sources & references

1. [Robota-Embed: Unified VLA Embeddings for Robot In-Context Learning and Evolution](https://arxiv.org/abs/2610.19495) – arXiv cs.RO, 2026-10-15

Last updated: 2026-10-15