Generalist robot policies are trained on static offline datasets, yet physical manipulation unfolds sequentially across stages. Because a weakness in a downstream stage remains unobservable until the policy passes earlier ones, each resolved bottleneck exposes the next hidden failure mode. This staged structure calls for Recursive Self-Improvement (RSI), in which a policy iteratively executes, diagnoses failures, and updates itself. Moreover, because task completion behaves like a product of stage-level pass rates, recursive gains can multiply across generations. In physical environments, three bottlenecks obstruct compounding: terminal execution feedback rarely localizes the stage that caused a failure, failed trajectories do not supply the corrective action, and updating shared parameters can degrade previously mastered skills. We propose Compounding Physical RSI, a self-evolving policy framework that turns execution failures into heritable capability. First, automated capability-gap assessment traces failed rollouts to the task stage where progression breaks down and formulates a structured improvement objective. Second, conditioned on this objective, our remedy stage curates validated, stage-relevant experience and refines the policy via behavior-regularized offline reinforcement learning, without requiring costly online rollouts at each optimization step. Third, to safeguard existing skills, our retention stage constructs weight-space candidates between the inherited reference policy and the refined policy, evaluates them across the task suite, and promotes the best candidate as the reference state for the next generation. Evaluated across all 42 tasks of the RoboDojo-Sim benchmark, CSU-AI-07 ranks first on the leaderboard with a three-seed mean Overall Score of 26.15 and Success Rate of 20.70%, outperforming the next-best entry by 5.92 points and 5.82 percentage points.
Figures & tables
Figure 1: Architecture of Compounding Physical RSI. Closed-loop execution in RoboDojo-Sim feeds automated stage-level capability-gap assessment (Diagnosis), which produces a structured objective gk to guide experience curation and offline reinforcement learning (Remedy). Retention interpolates candidates between the inherited reference policy and the refined policy, promotes the best-performing candidate via full-benchmark evaluation, and passes it forward as the reference state for the next generation.
Figure 2: Diagnosis and remedy across two representative manipulation failure modes. (a) Contact-to-transition failure on a desk object, where physical contact fails to advance the task stage. (b) In-hand transport instability, where an object is lifted successfully but slips during transport. In both cases, Diagnosis traces execution evidence to the specific failing stage, conditioning the selection of validated corrective experience in Remedy.
Evaluation
Overall Score ↑
Success Rate (%) ↑
Seed 0
26.53
21.05
Seed 1
25.46
20.31
Seed 2
26.45
20.75
Three-seed mean
26.15
20.70
Table 1: Full-benchmark RoboDojo-Sim evaluation results for CSU-AI-07 across three independent evaluation seeds.
Policy
Overall Score ↑
Success Rate (%) ↑
CSU-AI-07
26.15
20.70
GalaxeaVLA (G0.5) [ 36 ]
20.23
14.88
Xiaomi-Robotics-1 [ 11 ]
20.07
13.93
Hy-Embodied-0.5-VLA [ 37 ]
13.07
8.80
SimpleMemVLA [ 38 ]
12.58
9.27
Spatial Forcing [ 9 ]
12.38
8.04
Table 2: Overall RoboDojo-Sim leaderboard comparing CSU-AI-07 against thirteen public generalist policies. The benchmark protocol follows RoboDojo-Sim [ 34 ] .
Robots are increasingly autonomous in task execution, yet their improvement remains largely human-driven. This study investigates Physical Recursive Self-Improvement (Physical RSI), in which physical experience guides lasting updates to both a robot's capabilities and its improvement procedure, making subsequent improvement more effective across tasks. Physical RSI faces challenges from noisy observations, continuous dynamics, costly trials, and irreversible consequences. We present an L0-L5 framework that differentiates capability-level from meta-level updates and within-task from across-task gains, reserving L5 for independently verified recursive gain. We study existing methods around harness adaptation and model training, alongside safety and control mechanisms that govern execution and successive updates. Our analysis reveals that, despite substantial progress in self-improvement, the surveyed literature does not yet demonstrate Physical RSI as defined by our framework. We identify research priorities in reliable feedback, causal attribution, safety preservation, and benchmarking, outlining a path toward robots that autonomously improve how they learn while ensuring safety. The GitHub repository for Physical RSI is available at https://github.com/SAIL-Research-Lab/Physical-RSI.
Hai Zhou, Fan Yang, Zeyi Chen +27
Shanghai Jiao Tong University · Google · Microsoft +13
Recursive self-improvement (RSI) describes a system that improves not only its behavior but also the process by which it improves. Recent digital agents test this idea by rewriting their own code and checking each version with executable tests. Bringing these mechanisms into the physical world requires learning from partial observations through costly trials, while accounting for changing operating conditions and potentially irreversible consequences. We position RSI within robot learning and trace pathways from improving robot capabilities to improving the processes that generate them. We examine how physical experience can strengthen these improvement mechanisms and when the resulting gains can support further self-improvement of the mechanisms themselves. Gains carry forward only when feedback addresses remaining learning needs, the receiving component can use the change, its value survives changes in tasks or hardware, and continued practice remains safe and affordable. Existing studies, concentrated in manipulation, support behavioral improvement, benefits to later learning, and procedure improvement. Some systems also change the improver's knowledge or tools, but the physical studies reviewed here have yet to isolate gains attributable to recursion. We synthesize these findings into evaluation principles and research directions for machines that, trial by trial, safely improve the way they improve.
Meng Luo, Zechen Bai, Shanqing Xu +9
National University of Singapore · Huazhong University of Science and Technology · Shanghai Jiao Tong University +2
Embodied systems can update policies, world models, skills, and executable programs from deployment experience, allowing one interaction to benefit later tasks. When a system also revises and reuses its data-acquisition, feedback, or learning procedures, experience changes the process that produces capability, introducing recursive self-improvement. This paper surveys research in physical AI through the connection between persistent capability and its improvement process. We first explain the functional distinction between learner updates and improvers, then compare policy learning, imagined experience, memory and skill accumulation, and executable harnesses. These approaches connect to revisions of curricula, rewards, and research programs in different ways. Contact validity, recovery, retention, and resource costs determine whether the resulting gains accumulate during continued deployment. An L0-L3 rubric distinguishes task-local adaptation, persistent capability gain, inherited process revision, and verified improver gain. Representative benchmarks and within-study contrasts relate these distinctions to existing results. We further propose common-data and interactive comparisons that evaluate reference and revised improvers from shared learner starts, separating the use of experience from its acquisition. This perspective connects the diverse methods, physical operating conditions, and evaluation evidence, underpinning embodied RSI. A comprehensive list of studies in this survey is available at https://github.com/nkd-lkz/physical-ai-rsi-survey.
Xiaofeng Han, Kaize Luo, Zenghuang Fu +6
School of Artificial Intelligence, University of Chinese Academy of Sciences, China · The State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, China · Southern University of Science and Technology, China +3