ME-Brain-1.0: Memory, Cognition and Action for Evolving Embodied Intelligence
cs.RO
Submitted: 2026-09-21
Updated: 2026-09-29
Code: https://github.com/MachEmbodied/ME-Brain-1.0
Project page: https://machembodied.com/ME-Brain/ME-Brain-1.0.html
License: http://creativecommons.org/licenses/by/4.0/
The gist: Current embodied systems largely rely on pretrained capabilities that remain fixed after deployment, limiting their ability to learn from physical interaction.
Terminology
Abstract
Current embodied systems largely rely on pretrained capabilities that remain fixed after deployment, limiting their ability to learn from physical interaction. We introduce MachEmbodied-Brain (ME-Brain), a self-evolving embodied system organized around a closed loop of action execution, experience acquisition, experience evolution, and improved execution. Evolvable Memory consolidates multimodal trajectories into hierarchical, reusable experience; Cognitive Core transforms physical experience into transferable skills; and the Action Model combines event-driven keyframes, EventCell local-world prediction, and action-conditioned memory modulation to focus computation on decision-critical moments, regions, and historical evidence. Together, these modules shift embodied intelligence from train-and-freeze to deploy-and-evolve without model retraining. Cognitive Core outperforms the strongest comparison models by 8.2 and 9.6 points on embodied and agent benchmarks. The Action Model achieves 47.88% mean success on RoboMME, a 3.26-point improvement over the strongest baseline. On RoboDojo, it reaches a 21.51 mean Score and 16.03% success rate, exceeding π 0.5 by 10.10 and 9.12 points. On the six-task ME-RealBench, ME-Brain achieves a 69.5 mean Score and 66.7% success rate, outperforming DM0.5 by 12.8 and 11.7 points, respectively.
Sources
- PaliGemma: A versatile 3B VLM for transfer
- OpenVLA: An Open-Source Vision-Language-Action Model
- $\pi_0$: A Vision-Language-Action Flow Model for General Robot Control
- $\pi_{0.5}$: a Vision-Language-Action Model with Open-World Generalization
- RoboOS: A Hierarchical Embodied Framework for Cross-Embodiment and Multi-Agent Collaboration
- PhyAgentOS: A Self-Evolving Operating System for Embodied Agents with Decoupled Cognitive Planning and Physical Execution
- ACE-Brain-0.5: A Unified Embodied Foundational Model for Physical Agentic AI
- EmbodiSkill: Skill-Aware Reflection for Self-Evolving Embodied Agents
- Robots Acquire Manipulation Skills in Seconds from a Single Human Video
- Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory
- MemOS: A Memory OS for AI System
- MemoryVLA: Perceptual-Cognitive Memory in Vision-Language-Action Models for Robotic Manipulation
- MemER: Scaling Up Memory for Robot Control via Experience Retrieval
- MEM: Multi-Scale Embodied Memory for Vision Language Action Models
- RoboClaw: An Agentic Framework for Scalable Long-Horizon Robotic Tasks
- World Action Models are Zero-shot Policies
- GigaWorld-Policy: An Efficient Action-Centered World--Action Model
- EventVLA: Event-Driven Visual Evidence Memory for Long-Horizon Vision-Language-Action Policies
- Infrastructure First: Enabling Embodied AI for Science in the Global South
- Kairos: A Regret-Aware Native World-Action Model Stack for Physical AI
Related papers
- FMT x: An Efficient and Asymptotically Optimal Extension of the Fast Marching Tree for Dynamic Replanning
- MPCFormer: A physics-informed data-driven approach for explainable socially-aware autonomous driving
- RoboLab: A High-Fidelity Simulation Benchmark for Analysis of Task Generalist Policies
- ARC: A Reasoning Recipe for Robot Foundation Models
- HRDexDB: A 4D Dexterous Grasping Dataset Across Human and Multiple Robot Embodiments
- APT: Action Expert Pretraining Improves Instruction Generalization of Vision-Language-Action Policies