META-Robot: Self-Cognition and Capability Scaling in Agentic Robots
cs.RO
Submitted: 2026-10-30
Updated: 2026-11-04
Code: https://github.com/deepseek-ai/deepseek-harness
License: http://creativecommons.org/licenses/by/4.0/
The gist: LLM-based robotic agent systems can generate programs for robot control, yet their generalization to new tasks often depends on elaborate scaffolding.
Terminology
Abstract
LLM-based robotic agent systems can generate programs for robot control, yet their generalization to new tasks often depends on elaborate scaffolding. Expanding their skill repertoires alone offers limited guidance on how to assess new capabilities or identify what remains beyond their reach. We introduce META-Robot, a framework that couples self cognition with capability scaling by expanding the criteria through which an agent evaluates and develops its skills. At its core, action-level predicates make execution outcomes explicit and available for LLM reasoning. The agent uses these assessments to identify its own limitations, extend relevant evaluation criteria, and guide the structured expansion of its skills. Proprioceptive feedback further grounds this process in the robot's physical state, enriching the evidence available for self-cognition. Experiments in simulation and on real robots demonstrate stronger overall task performance, generalization to new tasks, continued expansion of skills and evaluation criteria, and effective use of non-visual proprioceptive signals.
Sources
- $\pi_0$: A Vision-Language-Action Flow Model for General Robot Control
- RHO: Your Coding Agent is Secretly a Roboticist
- CaP-X: A Framework for Benchmarking and Improving Coding Agents for Robot Manipulation
- $\pi_{0.5}$: a Vision-Language-Action Model with Open-World Generalization
- Memory Transfer Planning: LLM-driven Context-Aware Code Adaptation for Robot Manipulation
- Act-Observe-Rewrite: Multimodal Coding Agents as In-Context Policy Learners for Robot Manipulation
- Guava: Distilling Frontier VLMs into a Compact Agent through a Robotic Manipulation Harness
- ASPIRE: Agentic /Skills Discovery for Robotics
- Robotic Ultra-Long-Horizon Manipulation Skills via Human-guided Lifelong Code Generation
- Robotic Long-Horizon Manipulation with Progressive In-Context Code Generation and Episodic Feedback
- ALRM: Agentic LLM for Robotic Manipulation
- Hi Robot: Open-Ended Instruction Following with Hierarchical Vision-Language-Action Models
- Evolvable Embodied Agent for Robotic Manipulation via Long Short-Term Reflection and Optimization
- EmbodiedSkills: World Action Model-augmented Cascaded Skills for Vision-Language-Action Agent
- ENPIRE: Agentic Robot Policy Self-Improvement in the Real World
- DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
- Being-0: A Humanoid Robotic Agent with Vision-Language Models and Modular Skills
- Playful Agentic Robot Learning
- LIBERO-PRO: Towards Robust and Fair Evaluation of Vision-Language-Action Models Beyond Memorization
Related papers
- Rethinking Actions: Continuity-Aware Priors for World-Model Planning
- MOSAIC: Modular Obstacle and Spatial-Aware Framework for Manipulation In Cluttered Scenes
- SAGAS: Semantic-Aware Graph-Assisted Stitching for Offline Temporal Logic Planning
- DexSeed: Expanding Human Hand-Object Interactions into Robot Demonstrations for Dexterous Policy Learning
- LaST0.5: A Generalist World Action Model with Interaction-Centric Latent Reasoning
- CIALLO: Learning Shared Intent Representations for Cross-Embodiment Manipulation