Skip to content
AI News HubLIVE
More
Summary only

Embodied AI 'R1 Moment': LaST-R1, the LIBERO Terminator, Sets New Paradigm with 99.9% via Physical Reasoning

LaST-R1 introduces latent-space physical reasoning into reinforcement learning for robotic manipulation, achieving 99.9% average success on LIBERO with just one demonstration trajectory and outperforming π0.5 by 22.5% on real-world tasks.

Source量子位Author: 思邈
Embodied AI 'R1 Moment': LaST-R1, the LIBERO Terminator, Sets New Paradigm with 99.9% via Physical Reasoning
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions

Summary only

Under the source content permission, this page provides an editorial summary, key points, and a link to the original article.

Key points and analysis

Article intelligence

EngineersIntermediate

Key points

  • LaST-R1 performs latent Chain-of-Thought reasoning before action generation, enabling robots to understand physical dynamics.
  • The LAPO algorithm jointly optimizes latent reasoning and action policies using environment rewards.
  • Achieves 99.9% success on LIBERO and 93.75% on real-world tasks with minimal training data.
  • Demonstrates robust generalization to unseen objects, backgrounds, and lighting conditions.

Highlights and analysis are generated automatically and may contain errors. Check the original source.

More recent stories on this topic · Agents