arXiv AI

EmbodiedRSI: Active Continual Robot Learning Through Hypothesis-Guided Co-Evolution

EmbodiedRSI is a self‑evolving agentic harness that autonomously selects physical experiments to refine robot code and skills, using a Fast‑Slow Dual‑System Architecture and a Hypothesis Graph. It employs Value‑of‑Information Experiment Selection to distinguish competing hypotheses, guiding Code‑Skill Co‑Evolution, while the Slow System builds Hierarchical Memory and Reward‑Grounded Memory Learning to improve future Fast‑System performance. On RoboCasa365 and LIBERO‑Pro benchmarks, EmbodiedRSI achieves markedly higher success rates than baselines and demonstrates zero‑shot transfer to real‑world robots, achieving 71.3% overall success across multiple challenging tasks.

arXiv AI
Jul 2

ASPIRE: Agentic /Skills Discovery for Robotics

arXiv:2607. 00272v1 Announce Type: cross Abstract: Traditional robot programming is challenging: it requires orchestrating multimodal perception, managing physical contact dynamics, and handling diverse configurations and execution failures.

By Runyu Lu, Yubo Wu, Ethan Kou, Letian Fu, Wenli Xiao, Ajay Mandlekar, Yinzhen Xu, Guanya Shi, Ken Goldberg, Ang Chen, Mosharaf Chowdhury, Yuke Zhu, Linxi "Jim" Fan, Guanzhi Wang
arXiv AI
Jun 19

Playful Agentic Robot Learning

arXiv:2606. 19419v1 Announce Type: cross Abstract: Current agentic robot systems can write executable Code-as-Policy programs, observe feedback, and revise behavior across multiple attempts, but they remain largely task-driven: reusable skills are acquired only after explicit instructions.

By Junyi Zhang, Jiaxin Ge, Hanjun Yoo, Letian Fu, Zihan Yang, Yaowei Liu, Raj Saravanan, Shaofeng Yin, Justin Yu, Dantong Niu, Zirui Wang, Roei Herzig, Ken Goldberg, Yutong Bai, David M. Chan, Ion Stoica, Angjoo Kanazawa, Jiahui Lei, Haiwen Feng, Trevor Darrell
arXiv Computer Vision
Nov 2

RoboArchitect: Automating the Design of Robotic Agents with Persistent Memory

RoboArchitect automates the design of agent harnesses that orchestrate frozen robot policies, vision tools, and direct arm commands. By observing rollouts, diagnosing failures, and iteratively revising the harness, it improves success rates from 9.7% to 43.3% on unseen scenes and from 12.7% to 40.2% on unseen tasks without updating model weights. The evolved harness also develops persistent memory and intuitive control strategies, outperforming finetuned policies especially when the benchmark diverges from the policy’s training data.

By Shengguang Wu, Suning Huang, Cherie Ho, Ria Doshi, Hao Zhu, Serena Yeung-Levy
arXiv AI
Oct 12

EmbodiedCLBench: Evaluating Continual Learning for Self-Evolving Embodied Agents

EmbodiedCLBench is the first benchmark designed to evaluate continual learning in embodied simulation environments, focusing on self‑evolving agents. The benchmark uses compositional generalization to test whether agents can learn basic tasks and then recombine those skills to solve more complex, unseen tasks. Experiments across various agent harnesses and learning methods reveal that while basic task experience improves performance on advanced tasks, this benefit does not scale with more learning samples and current agents fail to extract additional useful information.

By Jie Huang, Yanan Chen, Ruixun Liu, Kaichen He, Xinyi He, Rui Huang, Zilong Zheng, Zhenliang Zhang, Yiwu Zhong
arXiv AI
Nov 6

Learning the Task or Learning the Verifier? Rethinking Criteria Exposure in Embodied Self-Improvement

The paper investigates the impact of exposing self‑improving embodied agents to benchmark success criteria, a practice termed Criteria Exposure (CE). Experiments show that while CE leads to higher success rates on the exposed criteria, the resulting code overfits and performs poorly when the criteria are slightly altered or when transferred to new tasks. The authors propose instruction‑faithful criteria that more accurately reflect task completion by agent‑generated programs.

By Wenbo Huang, Jianxin Bi, Shunyu Yang, Ruiyang Si, Rui Ni, Qiang Wang, Shulong Jiang, Yintianrun Zhang, Feixiang Ruan, Kewei Zhang, Ye Huang, Shukai Gong, Daquan Zhou
arXiv Computation and Language
Sep 11

Self-Evolving Embodied Agents via Skill-Harness Evolution

The paper introduces SHAPER, a self‑evolving framework that enables embodied agents to adapt to new environments without retraining the underlying foundation model. SHAPER keeps model parameters frozen and instead evolves reusable skills and a context‑code harness through rollouts in the target environment, allowing the same model to act as both planner and optimizer. Experiments on VLABench and ESI‑Bench demonstrate that this skill‑and‑harness optimization outperforms pure execution, supervised fine‑tuning, and test‑time scaling baselines, showing that self‑evolving adaptation is viable when model training is costly or impractical.

By Peidong Wang, Zhiming Ma, Ying Chang, Xufang Luo, Yiqun Zhang, Zihan Wang, Xiaocui Yang, Shi Feng, Yuqing Yang, Dongsheng Li
arXiv Computation and Language
Oct 28

OmniBrain-0: A Unified Agentic Model for Embodied Intelligence

OmniBrain-0 is a lightweight, unified embodied agent that coordinates manipulation and navigation through a shared high‑level decision model. It integrates task understanding, planning, progress tracking, and feedback‑driven recovery, and is trained via embodied pretraining, domain‑specialized agent distillation, and multi‑teacher on‑policy distillation. The framework also introduces Rubric‑Guided Recursive Self‑Improvement, a data flywheel that refines rubrics and trajectories for continual domain‑expert updates, enabling efficient transfer and consolidation of embodied capabilities across multiple benchmarks and real‑world tasks.

By Xiaomi Embodied Intelligence Team, Yuechen Luo (University of Macau, Tsinghua University), Chong He (University of Macau, Tsinghua University), Fang Li (University of Macau, Tsinghua University), Shaoqing Xu (University of Macau, Tsinghua University), Jingkai Wang (University of Macau, Tsinghua University), Menglun Su (University of Macau, Tsinghua University), Jiaxin Liu (University of Macau, Tsinghua University), Shengyin Jiang (University of Macau, Tsinghua University), Zhiyi Lai (University of Macau, Tsinghua University), Long Chen (University of Macau, Tsinghua University), Longlong Wang (University of Macau, Tsinghua University), Liang Hu (University of Macau, Tsinghua University), Haochen Liu (University of Macau, Tsinghua University), Yuncheng Jiang (University of Macau, Tsinghua University), Hongwei Xie (University of Macau, Tsinghua University), Jieming Cui (University of Macau, Tsinghua University), Hanbing Li (University of Macau, Tsinghua University), Junjie Huang (University of Macau, Tsinghua University), Zhixiang Duan (University of Macau, Tsinghua University), Hexin Zhang (University of Macau, Tsinghua University), Weixiang Liang (University of Macau, Tsinghua University), Ying Gong (University of Macau, Tsinghua University), Yuhan Wang (University of Macau, Tsinghua University), Yong Pan (University of Macau, Tsinghua University), Pengna Li (University of Macau, Tsinghua University), Kangyi Wu (University of Macau, Tsinghua University), Ken Deng (University of Macau, Tsinghua University), Kun Ma (University of Macau, Tsinghua University), Fuxi Wen (University of Macau, Tsinghua University), Zhi-xin Yang (University of Macau, Tsinghua University), Hangjun Ye (University of Macau, Tsinghua University)
arXiv Machine Learning
1d ago

Training Reasoning Models in Simulation to Natively Use Real Robots

The paper demonstrates that an open‑weight 27B reasoning model can be trained entirely in simulation to natively control real robots using reinforcement learning, without demonstrations. By employing a Post‑training Engine for Agentic Robotics (PEAR) and robot‑use reinforcement learning (RURL), the model learns to observe, reason, and generate end‑effector poses, achieving 60% success on a real dual‑arm robot after training on simulated tasks. The approach shows strong generalization to unseen robots, real‑world settings, and novel instructions, while preserving performance on reasoning, coding, and vision benchmarks.

By Jiaming Tang, Zhuoyang Zhang, Junxian Guo, Fangzhou Zhao, Yilong Zhao, Zekai Li, Xingyang Li, Yuyang Zhao, Song Bian, Zhijian Liu, Song Han
arXiv AI
Aug 19

Teach and Grow: An Agent-Centered Architecture for General Robot Learning

Teach-and-Grow Learning (TGL) is an agent-centered architecture that transforms a few successful demonstrations into reusable Skill Blocks, enabling a robot to compose, execute, and revise behaviors in new scenes without task-specific policy retraining. The system maintains a Skill Library and structured Experience Memory to capture successes, failures, and repairs, allowing persistent reuse and agent-directed adaptation. Evaluation on the LIBERO benchmark shows state-of-the-art performance, and the authors propose a scaling-law hypothesis suggesting that accumulated reusable experience reduces future-task error and teaching demand following a power-law trend.

By Chang Nie, Zhe Liu, Hesheng Wang