Computation and Language · arXiv:2610.20474 v1 · submitted 14 Oct 2026 · 39 references
AdaRSI: On-Demand Recursive Self-Improvement for Embodied Agents
The paper says
“Pretrained vision-language-action models (VLAs) often fail midway through unseen long-horizon tasks.”
more of the abstract
Recursive self-improvement (RSI) offers a remedy, but embodied RSI methods either refine the harness around a frozen executor and so cannot add a missing motion, or train the executor on persistent failures without asking whether a failure lies in how the executor is called or in what it can do. We ask when a self-improving embodied system should change the weights of its executor, and propose AdaRSI, an on-demand RSI framework for embodied agents. A coding agent first improves its skills and harness, then attributes persistent failures by retrying failing sub-steps from saved states, and fine-tunes only the sub-steps that remain unreliable, with demonstrations it collects itself and experience carried over from earlier tasks. On all 50 target tasks of RoboCasa365, with GR00T N1.5 as the executor and no human demonstrations, AdaRSI improves the VLA on all three suites and raises its success on unseen composite tasks from 4.6% to 13.8%. Compared with fine-tuning every sub-step, it reaches slightly higher success while using only 28% of the GPU time and 22% of the self-collected demonstrations, and fine-tuning without a harness trains more yet gains far less.
from the abstract
Pith says
Machine review · deepseek-v4-flash · 15 Oct 2026 · what was read was not recorded · not checked by a person
It can be wrong. Challenge this review or reply to it.
Holds up, with conditions. A sound empirical answer to when an embodied system should retrain its executor, with a genuinely new attribution mechanism and a strong control—but the headline numbers pool uncapped pilot tasks with budget-limited ones and need re-separation. This line is wrong
3 major objections · 6 minor
How often objections from reviews like this one are right: not yet measured
Not checked: figures, code, what the 39 references say.
The paper says, at §4.3, §5.1, Tables 1, 2, 7
The review did not quote this passage.
Pith says: Objection 1 of 3 · major
The pilot-task protocol confound is real and load-bearing. Section 4.3 states that pilot tasks are 'improved without any limit on training,' §5.1 states that pilots exist in each suite, and Tables 1 and 7 pool all 50 tasks (respectively all 16 Composite-Unseen tasks) into the headline numbers; at least GarnishPancake and HeatKebabSandwich are identifiable as Composite-Unseen pilots from Table 9, and the pilots were selected on development data to span base-VLA success levels.
more of objection 1
The headline claims (48.8 vs 42.4 overall, 13.8 vs 4.6 on Composite-Unseen, 28% GPU time and 22% demonstrations vs Always train) therefore mix budget-limited non-pilot tasks with uncapped pilot tasks, and the pilot set is never enumerated. To the authors' credit, the two identifiable Composite-Unseen pilots contribute no successes in Table 2, so the confound's magnitude is unclear rather than certainly harmful, and the non-pilot breakout in Table 8 (harness 12.0% to AdaRSI 16.6%) suggests the on-demand result survives for that suite. Nevertheless, this must be fixed before the central claim can be verified: enumerate the pilot tasks per suite, and re-report the results of Tables 1, 2, and 7, including the 3.5×/4.5× cost ratios, on non-pilot tasks alone or side-by-side with the pooled numbers.
The paper says, at §5.2, §5.3, Tables 1, 5, 8
The review did not quote this passage.
Pith says: Objection 2 of 3 · major
Suite-level and task-level comparisons are reported as point estimates with no confidence intervals or paired tests. Several load-bearing contrasts are small relative to 50-scene-per-task sampling noise: the weight-round increments on Atomic-Seen (+1.4 pp) and Composite-Seen (+1.0 pp) in Table 1, the 'slightly higher success' of AdaRSI over Always train (13.8 vs 12.6 in Table 5), and the lowest-first versus harness comparison (11.8 vs 12.0 in Table 8).
more of objection 2
Please report per-task paired differences with intervals (bootstrap or exact binomial tests), so the reader can see which of the small suite-level increments are distinguishable from noise; per-task tables such as Table 2 would also benefit from flags on differences of a few points that are within noise.
The paper says, at §4.2, §6, Table 8
The review did not quote this passage.
Pith says: Objection 3 of 3 · major
The definition of an executor-limited sub-step as one for which 'no harness the agent can find' lets the executor perform it reliably folds the coding agent's search capability into the ground truth. The paper explicitly acknowledges this dependence in Sec. 6, and the lowest-first ablation in Table 8 is the right kind of external control, showing that selecting the same sub-steps without retrying from other states or instructions gains nothing.
more of objection 3
What remains unresolved is quantitative: the paper never reports how often the retry probe actually found a working instruction or approach (that is, how often attribution correctly redirected effort back to the harness), nor the number of probe attempts per trained sub-step. Please report these probe statistics and, if feasible, the sensitivity of the chosen training set to the probe budget; without this, the reader cannot tell how often the weight round trained a failure that was in fact harness-callable by a stronger search.
6 minor objections
The paper says, at §5.3, Table 8
Pith says: minor objection 1 of 6
Report how many tasks in the 'lowest first' arm actually trained at least one policy; with four tasks exhausting the development session before any policy was trained, the comparison partly measures failed sessions rather than selection quality alone.
The paper says, at §5.3, Table 7
Pith says: minor objection 2 of 6
The statement that the counts of Always train include the policies of AdaRSI, which it reuses, should appear in the cost comparison itself; since AdaRSI's 966 GPU minutes are counted in both columns, the marginal cost of the extra training is about 2.5× (3,411 minus 966 minutes), so the reported 3.5× is conservative and should be labeled as such.
show 4 more
The paper says, at §5.1
Pith says: minor objection 3 of 6
Clarify that 'the official protocol' refers to the evaluation protocol (target split, task horizons, success check, 50 test scenes), because the improvement phase, with privileged snapshots, saved states, and uncapped pilot training, is not part of that protocol and should be described as an additional development procedure.
The paper says, at §4.2, §4.3
Pith says: minor objection 4 of 6
Specify the two rules that currently drive the whole pipeline: the reliability threshold for declaring a sub-step executor-limited (and for adopting a policy), and the fixed rule that maps pilot training logs to the per-task training budget; both are described only qualitatively, which limits reproducibility.
The paper says, at §5.4, Figure 4
Pith says: minor objection 5 of 6
The 'training cost' axis counts GPU minutes and demonstrations but excludes the language-model tokens of the improvement phase, which are large (154M input tokens per Composite-Unseen task); state explicitly in the figure caption and the cost section that the comparison is in training compute, with LM-token costs reported separately.
The paper says, at §5.1
Pith says: minor objection 6 of 6
Document the exact model versions and decoding settings for the coding agent (GPT-6.1-sol via Codex) and the two GR00T N1.5 checkpoints, to aid reproducibility of the agent-harness component.
What would settle it
Publish the pilot-task list per suite and recompute Table 1, Table 5 and Table 7 restricted to the non-pilot tasks, the same operation already performed for Table 8 on Composite-Unseen. If the suite-level gains and the cost ratios persist outside the pilot tasks, the headline stands; if the gains concentrate in the pilot tasks, the abstract's "on all 50 target tasks" claim must be revised to non-pilot tasks or re-evaluated under a single budget-limited protocol.
from the skeptic's reading
How the review went
4 readings by deepseek-v4-flash · 15 Oct 2026
- Reader holds up, with conditions
- Skeptic partly agrees
- Editor wrote the verdict above
- Referee major revision: 3 major objections, 6 minor
Not run on this paper: the simulated authors' defence.
Pith did not check this review's objections against the paper.
The review in full
The review in brief, machine-written
The paper asks the right question—when should a self-improving embodied system change the executor's weights instead of the harness around it—and answers with a mechanism that is actually new: save state, retry the failing sub-step with different instructions and approaches, and train only if it stays unreliable. The "lowest first" ablation is the strongest evidence: same budget, same harness, same number of trained sub-steps, no attribution, and it gains nothing over the harness alone. That control keeps the on-demand claim from being circular.
The experiments are mostly careful. Official RoboCasa365 protocol, disjoint scenes, held-out starts, paired sub-step tests from identical states, and cost accounting that includes trained-but-unadopted policies. The paired tests—fine-tuning teaches missing motions like the microwave door, degrades motions pi0 already completes—are the cleanest evidence, and the paper honestly reports the CuttingToolSelection regression and discusses it in Section 6.
The stress-test concern checks out. Section 4.3 says pilots run without any training limit and their logs set the budget; Section 5.1 confirms pilots are part of the 50 tasks; Table 1 pools them. So the headline 48.8% vs 42.4%, and the 28%/22% cost ratios, mix uncapped pilots with budget-limited tasks, and the pilots were chosen to span easy-to-hard, not randomly. The only non-pilot breakout, Table 8 on Composite-Unseen, is reassuring (16.6% vs 12.0% for the harness alone, and the budget-matched lowest-first rule gains nothing), but there are no equivalents for the other suites or the cost comparison. This must be fixed; I don't think it is fatal.
Smaller issues: no confidence intervals or paired tests anywhere, on per-task counts that are often 0–7 out of 50; no code, prompts, or logs; and the harness is written by a proprietary model (GPT-6.1-sol via Codex), so the pipeline is only partly reproducible. The attribution caveat—"executor-limited" is defined relative to what this agent can find—is real, but the authors state it in Section 6, and the lowest-first control gives it teeth.
For people working on VLA adaptation, agentic harnesses, and self-improving robots, this is worth a serious referee. The question is well posed, the non-pilot evidence supports the principle, and the fixes are clear: list the pilot tasks, re-report all main numbers with pilots excluded, add intervals, and release the harness and logs. Send it to review.
The strongest objection
Section 4.3 says "The pilot tasks are improved without any limit on training" and that their logs set the experience document and training budget for later tasks; Section 5.1 confirms pilots exist in each suite and are among the 50 target tasks summarized in Table 1. The headline comparisons—48.8% vs 42.4% overall, 13.8% vs 4.6% on Composite-Unseen, and 28% GPU / 22% demos relative to always-training—therefore pool budget-limited non-pilot tasks with uncapped pilot tasks. The pilot set is not listed, and pilots were selected by the authors on development data to span high, moderate and low base-VLA success (§4.3), so they are not a random subset. A reader cannot tell how much of the gain comes from tasks that received unlimited training and whose logs shaped the experience document. The paper isolates non-pilot Composite-Unseen in Table 8 (Agent+harness 12.0% to AdaRSI 16.6%), which suggests the on-demand result survives there, but no equivalent non-pilot breakouts are given for Atomic-Seen, Composite-Seen, or the cost comparison in Table 7. This is an evaluation-protocol confound in the headline empirical claim, not merely the acknowledged attribution limitation of Sec. 6: even with perfect attribution, the "all 50" claim is unverifiable until pilot and non-pilot results are separated.
The reader's weakest_assumption is the attribution step: "executor-limited" is defined relative to the agent's own search (§4.2), so a weak search could mislabel harness-fixable failures as executor-limited. I agree this is real, but the paper already acknowledges it in §4.2 and §6, and all compared systems share the same agent search, so it bounds absolute performance rather than threatening the relative comparison against always-train or lowest-first. The more actionable weakness is the pilot-task protocol: §4.3 explicitly allows unlimited training on pilot tasks, §5.1 places pilots in every suite, and Table 1 reports all 50 tasks without separating them. If pilot tasks are excluded, the headline numbers may shift, and the cost advantage is part of the central claim. Table 8 provides one non-pilot breakout for Composite-Unseen and it supports the result, which is why I would not reject or downgrade the paper; the concern should be resolved by reporting the same breakout for all suites and for the cost tables. This matches the reader's conditional request to list pilot tasks, though it is not the reader's named weakest assumption, hence "partial" agreement. The verdict stays conditional pending that reporting.
The referee's report
If the reported results survive the pilot/non-pilot re-analysis, this is a solid empirical contribution: a clear, falsifiable design principle (attribute before training; harness first) instantiated on a standard benchmark with unusually careful evaluation hygiene. I credit the paper for testing on the official target split with disjoint development/validation/test scenes and held-out starts, for paired sub-step tests from identical states (Table 4) that directly measure what the weights change, for cost accounting that includes trained-but-unadopted policies, and for two genuinely informative external controls on the attribution claim (agent-without-harness and lowest-first under the same budget, Table 8). The case studies in Table 9 honestly document training failures, and Sec. 6 explicitly acknowledges the dependence of the method on the coding agent's capability. That said, the headline result is currently unverifiable as stated because pilot tasks trained without any budget limit are pooled into the main tables, and several load-bearing contrasts are reported without any uncertainty quantification. The paper's own limitation section does not mention the pilot-pooling issue.
Pith's reading of the paper (a machine's summary, not the authors')
When a long-horizon robot task fails midway, the failure itself does not say whether the robot was called the wrong way or simply cannot perform the required motion. AdaRSI's answer is to improve the calling layer first, an agent-written harness of tools, instructions, and procedures, and to change the executor's weights only for sub-steps that stay unreliable after every alternative way of calling has been tried. The paper's evidence is that this on-demand rule works: on all 50 target tasks of the RoboCasa365 benchmark, with a GR00T N1.5 vision-language-action model as executor and no human demonstrations, success improves on every suite, most sharply on unseen composite tasks, from 4.6% to 13.8%. Compared with fine-tuning every sub-step, on-demand training reaches slightly higher success while using 28% of the GPU time and 22% of the self-collected demonstrations. The load-bearing insight is that attribution, deciding whether a failure belongs to the harness or to the executor, is what makes training pay off, because it also determines the states from which a missing motion is learned.
What carries the argument. The mechanism that carries the argument is failure attribution by counterfactual re-execution. From saved simulator snapshots, the agent replays a failing sub-step from the same state under different instructions, poses, and approaches, and a sub-step that no such retry makes reliable is labelled executor-limited, separating a poor way of calling the executor from a motion the executor lacks. The harness that precedes this step is not merely a way to fix calling failures; it sets the states from which the executor is called, so a policy trained afterward learns the missing motion from the states in which it will actually be used. Validation completes the loop: every fine-tuned policy is compared with the base policy on held-out starts, and the procedure records recommendations so a new policy is called only where it is known to help, with the base policy kept as the default elsewhere.
Core claim. The paper's central claim is a principle: a self-improving embodied system should change the executor's weights only for a sub-step that stays unreliable after the agent has tried other ways of calling the executor, such as other instructions, robot poses, or approaches. It instantiates this as AdaRSI, a loop in which a coding agent first refines its harness, then attributes persistent failures by retrying failing sub-steps from saved simulator states, and finally fine-tunes only the executor-limited steps, those no harness the agent can find lets the base policy perform reliably, using demonstrations it collects itself. Each new policy is measured against the base policy from held-out states and is called only where it helps, so a weight change enters the system only after both attribution and validation. Empirically, on all 50 target tasks of RoboCasa365 with GR00T N1.5 as the executor and no human demonstrations, AdaRSI improves success on all three suites, from 4.6% to 13.8% on unseen composite tasks and from 75.4% to 79.7% and 43.0% to 49.0% on the seen suites, while reaching slightly higher success than fine-tuning every sub-step at 28% of the GPU time and 22% of the self-collected demonstrations.
The premise it rests on. The load-bearing premise is that the agent's search over instructions, poses, and approaches is a sound test of what the executor can do: a sub-step is judged executor-limited only if no harness this agent can find makes it reliable, so a failure to discover a working call is counted as a missing motion, a dependence on agent capability the paper itself notes in its discussion.
If the paper is right (Pith's inferences, not the authors')
Weight changes should be the last resort, not the default: fine-tuning every sub-step reaches lower success than on-demand training (12.6% vs 13.8% on Composite-Unseen) while using 3.5x the GPU time and 4.5x the demonstrations, and it can degrade sub-steps the executor already completes, as WaffleReheat falls from 38% to 28% when every sub-step is retrained.
Attribution decides not only which sub-steps to train but from which states to learn them: a rule that trains the lowest-success sub-steps first, under the same budget, gains nothing over the harness alone even though it selects mostly the same sub-steps (11 of 13).
The harness is the larger lever: on unseen composite tasks it more than doubles success, from 4.6% to 9.9%, before any training, and it makes later fine-tuning effective by fixing the poses and calls from which the executor operates.
Fine-tuning carries the tasks the harness cannot solve, such as opening a microwave door the base VLA never opens on its own (WaffleReheat, 0% to 38%), and there the trained weights decide whether the task succeeds.
Improvement compounds across rounds and tasks: an experience document and a training budget derived from pilot tasks guide later tasks, so the system improves not only itself but also the way it improves.
Beyond the paper: the executor-limited set is defined relative to the coding agent's search capability, so a stronger searcher should shrink it; a testable prediction is that shifting budget from fine-tuning toward the retry search would move the optimal harness/weight split toward the harness at the same total cost.
Beyond the paper: the same attribute-before-train rule applies to any system with a frozen policy and a separate caller, such as an LLM tool-user or a planner, where a failure's cause should be probed by changing the call before changing the model; the saved-state retry here is the embodied instance of that test.
Beyond the paper: the paired-state measurements suggest fine-tuning can degrade sub-steps that already work, so the method needs not only a gate that admits trained policies but a stopping rule for the retry search; monitoring how fast the retry success rate saturates with effort would provide one.
Check this review
Pith stored no record of this review's passes and took no fingerprint of it. Reviews made by code from 13 Nov 2026 on carry both.
How often reviews like this one are right
Reviews like this one: written by deepseek-v4-flash, where what was read was not recorded.
How often objections from reviews like this one are right: not yet measured
Reviews by people
This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.
The paper
Abstract
Pretrained vision-language-action models (VLAs) often fail midway through unseen long-horizon tasks. Recursive self-improvement (RSI) offers a remedy, but embodied RSI methods either refine the harness around a frozen executor and so cannot add a missing motion, or train the executor on persistent failures without asking whether a failure lies in how the executor is called or in what it can do. We ask when a self-improving embodied system should change the weights of its executor, and propose AdaRSI, an on-demand RSI framework for embodied agents. A coding agent first improves its skills and harness, then attributes persistent failures by retrying failing sub-steps from saved states, and fine-tunes only the sub-steps that remain unreliable, with demonstrations it collects itself and experience carried over from earlier tasks. On all 50 target tasks of RoboCasa365, with GR00T N1.5 as the executor and no human demonstrations, AdaRSI improves the VLA on all three suites and raises its success on unseen composite tasks from 4.6% to 13.8%. Compared with fine-tuning every sub-step, it reaches slightly higher success while using only 28% of the GPU time and 22% of the self-collected demonstrations, and fine-tuning without a harness trains more yet gains far less.
The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. No sentence in this record was written by a model.
abstract compiled · 2026-10-15
The authors' main claim, as the classifier picked it out 0.96
“On all 50 target tasks of RoboCasa365, with GR00T N1.5 as the executor and no human demonstrations, AdaRSI improves the VLA on all three suites and raises its success on unseen composite tasks from 4.6% to 13.8%.”
How the main claim is stated: asserted 1.00
What the abstract itself states, as probabilities: claims first 0.07 · states limitation 0.23 · human participants 0.04 · controlled comparison 0.94 · randomized assignment 0.02 · states quantitative result 0.99 · compares with prior results 0.95 · mentions code or data availability 0.05
The abstract, sentence by sentence
This table labels the abstract only (arXiv listing abstract v1). The number is the probability the classifier gave the label.
| Sentence | Role | The authors' own work | States the main conclusion | Strength |
|---|---|---|---|---|
| Pretrained vision-language-action models (VLAs) often fail midway through unseen long-horizon tasks. | background 0.99 | 0.09 | 0.02 | asserted 0.94 |
| Recursive self-improvement (RSI) offers a remedy, but embodied RSI methods either refine the harness around a frozen executor and so cannot add a missing motion, or train the executor on persistent failures without asking whether a failure lies in how the executor is called or in what it can do. | gap 0.87 | 0.07 | 0.04 | asserted 1.00 |
| We ask when a self-improving embodied system should change the weights of its executor, and propose AdaRSI, an on-demand RSI framework for embodied agents. | contribution 0.97 | 0.69 | 0.18 | asserted 0.77 |
| A coding agent first improves its skills and harness, then attributes persistent failures by retrying failing sub-steps from saved states, and fine-tunes only the sub-steps that remain unreliable, with demonstrations it collects itself and experience carried over from earlier tasks. | method 0.98 | 0.82 | 0.13 | asserted 0.99 |
| On all 50 target tasks of RoboCasa365, with GR00T N1.5 as the executor and no human demonstrations, AdaRSI improves the VLA on all three suites and raises its success on unseen composite tasks from 4.6% to 13.8%. | result 1.00 | 0.96 | 0.59 | asserted 1.00 |
| Compared with fine-tuning every sub-step, it reaches slightly higher success while using only 28% of the GPU time and 22% of the self-collected demonstrations, and fine-tuning without a harness trains more yet gains far less. | result 1.00 | 0.94 | 0.24 | asserted 1.00 |
What kind of paper the classifier reports this is: method or tool 1.00 · evidence from observation or experiment 0.94
What was examined, and what was not
Sources read
- arXiv listing abstract v1
Not examined
- full text
- figures and tables
- references and citing sentences
- relations to other papers
Levels run
- P abstract
Not run
- S sections
- T body sentences
- C citations
- R cross-paper relations
How the labels were made
Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-10-15: 1 request, 5,230 input tokens.
Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. This record as JSON
Figures (5)
Figures are the authors', reproduced from the paper.
References (39)
- [1] Kevin Black et al.π 0: A vision-language-action flow model for general robot control.arXiv preprint arXiv:2410.24164, on PithPith's reviewarXiv
- [2] Dian Chen, Brady Zhou, Vladlen Koltun, and Philipp Kr¨ahenb¨uhl. Learning by cheating. InConference on Robot Learning (CoRL), 2019. 4 on Pith
- [3] Mingguang Chen, Licheng Wang, and Bo Qu. Re- cursive self-improvement in AI: From bounded self- refinement to autonomous research loops.arXiv preprint arXiv:2607.07663, 2026. 2, 3 on PithPith's reviewarXiv
- [4] Yi Duan, Ying Liu, Zirui Tang, Haodong Chen, Jun Zhou, et al. The last AI built by humans: Toward genuine recursive self-improvement.arXiv preprint arXiv:2609.11873, 2026. 2, 3 on PithPith's reviewarXiv
- [5] Yuan Fang, Zechu Li, Haolei Tong, Puze Liu, and Georgia Chalvatzaki. Find something you can’t do: Agentic real- world reinforcement learning for self-improving VLA mod- els.arXiv preprint arXiv:2609.32069, 2026. 2, 3 on PithPith's reviewarXiv
- [6] L. Fu et al. CaP-X: A framework for benchmarking and im- proving coding agents for robot manipulation. InInterna- tional Conference on Machine Learning (ICML), 2026. 2, 3 on Pith
- [7] Hongcheng Gao, Jingjing Zhou, Zelin Zheng, Shijia Ge, Jay Zhu, Yazhe Wang, Jianshu Zeng, Xuan Shangguan, Di Wu, Lingyu He, Zhiqi Jia, Sihang Wu, and Xiao He. Self- evolving coding agents: From digital programs to physical- world intelligence.arXiv preprint arXiv:2609.35432, 2026. 2, 3 on PithPith's reviewarXiv
- [8] Jun Guo, Piaopiao Jin, Jason Li, Peiyan Li, Yingyan Li, et al. Xiaomi-Robotics-1: Scaling vision-language-action models with over 100k hours of real-world trajectories.arXiv preprint arXiv:2607.15330, 2026. 3 on PithPith's reviewarXiv
- [9] Linxuan He, Yuying Tian, Lingxiang Fan, Jiaqi Pi, Yinqiao Lu, Shang Su, Mengkai Shi, and Shuo Feng. Policy-level re- cursive self-improvement for embodied AI with a criticality world model.arXiv preprint arXiv:2607.28251, 2026. 2, 3 on PithPith's reviewarXiv
- [10] Puming Jiang, Tianrun Hu, Haozhe Du, Yibo Li, Zhiwei Xue, Xinhu Li, and Harold Soh. Agent priors-guided pol- icy learning.arXiv preprint arXiv:2609.35690, 2026. 2, 3 on PithPith's reviewarXiv
- [11] Moo Jin Kim, Karl Pertsch, Siddharth Karamcheti, et al. OpenVLA: An open-source vision-language-action model. InConference on Robot Learning (CoRL), 2024. 2, 3 on Pith
- [12] Haoran Lang, Haotao Lu, Shiyu Sang, Haoyang Luo, Guo Chen, Qun Li, Jingyi Yu, Ye Shi, and Jingya Wang. EmbodiRSI: Recursive self-improvement for data-efficient robot adaptation.arXiv preprint arXiv:2609.38905, 2026. 2, 3 on PithPith's reviewarXiv
- [13] Hyunin Lee, Jinglue Xu, Jeffrey Seely, Donghyun Lee, Matei Zaharia, and Yujin Tang. Recursive harness self- improvement.arXiv preprint arXiv:2607.15524, 2026. 3 on PithPith's reviewarXiv
- [14] Hao Li, Hangfan Zhang, Zhiyao Cui, Chunjiang Mu, Yiqun Zhang, Bo Zhang, Danyang Jia, and Shuyue Hu. RSI- Router: Evolving subtask-level LLM routing and skills for cost-efficient agents.arXiv preprint arXiv:2609.34712, on PithPith's reviewarXiv
- [15] J. Li, Y . Wang, H. Li, G. Zhang, Y . Ye, Q. Liu, H. Liang, and W. Lian. Practice makes policies: Bootstrapping and consol- idating robotic capabilities from zero human demonstrations. arXiv preprint arXiv:2607.26809, 2026. 2, 3 on PithPith's reviewarXiv
- [16] Jacky Liang, Wenlong Huang, Fei Xia, Peng Xu, Karol Hausman, Brian Ichter, Pete Florence, and Andy Zeng. Code as policies: Language model programs for embodied control. InIEEE International Conference on Robotics and Automa- tion (ICRA), 2023. 2, 3 on Pith
- [17] Jingsong Liang, Shuhao Liao, Shizhe Zhang, Diyuan Hou, Yuxin Cai, Xinjian Deng, Chengyang He, Wenhui Huang, Runjia Tan, Zhidong Wang, Lan Yu, Xuesong Tian, Guil- laume Sartoretti, Jie Luo, Yao Mu, Wenjun Wu, Wanhua Li, and Chen Lv. RoboFoundry: System-as-policy evo- lution for... on PithPith's reviewarXiv
- [18] R. Lu, Y . Wu, E. Kou, H. Fu, W. Xiao, A. Mandlekar, X. Xu, G. Shi, K. Goldberg, A. Chen, M. Chowdhury, Y . Zhu, L. J. Fan, and G. Wang. ASPIRE: Agentic skills discovery for robotics.arXiv preprint arXiv:2607.00272, 2026. 2, 3 on PithPith's reviewarXiv
- [19] Soroush Nasiriany, Abhiram Maddukuri, Lance Zhang, Adeet Parikh, Aaron Lo, Abhishek Joshi, Ajay Mandlekar, and Yuke Zhu. RoboCasa: Large-scale simulation of every- day tasks for generalist robots. InRobotics: Science and Systems (RSS), 2024. 2 on Pith
- [20] Soroush Nasiriany, Sepehr Nasiriany, Abhiram Maddukuri, and Yuke Zhu. RoboCasa365: A large-scale simulation framework for training and benchmarking generalist robots. InInternational Conference on Learning Representations (ICLR), 2026. 2, 3, 5 on Pith
show 19 more
- [21] NVIDIA. GR00T N1: An open foundation model for gen- eralist humanoid robots.arXiv preprint arXiv:2503.14734, on PithPith's reviewarXiv
- [22] Physical Intelligence, Kevin Black, Noah Brown, James Darpinian, Karan Dhabalia, et al.π 0.5: a vision-language- action model with open-world generalization.arXiv preprint arXiv:2504.16054, 2025. 3 on PithPith's reviewarXiv
- [23] Physical Intelligence, Bo Ai, Ali Amin, et al.π 0.7: a steer- able generalist robotic foundation model with emergent ca- pabilities.arXiv preprint arXiv:2604.15483, 2026. 2, 3 on PithPith's reviewarXiv
- [24] Lucy Xiaoyang Shi, Brian Ichter, Michael Equi, Liyiming Ke, Karl Pertsch, et al. Hi robot: Open-ended instruction 12 following with hierarchical vision-language-action models. InInternational Conference on Machine Learning (ICML), on Pith
- [25] Python Song, Zhixuan Liang, Kelsey Fu, Mengdi Wang, Jun- feng Yang, and Shilong Liu. EmbodiedRSI: Active contin- ual robot learning through hypothesis-guided co-evolution. arXiv preprint arXiv:2610.10498, 2026. 2, 3 on PithPith's reviewarXiv
- [26] Junyi Tang, Jie Peng, Zezhen Ding, Yuan Shen, and Tian- long Chen. AdaHVLA: Adaptive harnesses for long- horizon vision-language-action execution.arXiv preprint arXiv:2609.29204, 2026. 2, 3 on PithPith's reviewarXiv
- [27] Marcel Torne, Karl Pertsch, Homer Walke, Kyle Vedder, Suraj Nair, Brian Ichter, Allen Z. Ren, Haohuan Wang, Ji- aming Tang, Kyle Stachowicz, Karan Dhabalia, Michael Equi, Quan Vuong, Jost Tobias Springenberg, Sergey Levine, Chelsea Finn, and Danny Driess. MEM: Multi-scale embo... on PitharXiv
- [28] Vladimir Vapnik and Akshay Vashist. A new learning paradigm: Learning using privileged information.Neural Networks, 22(5–6):544–557, 2009. 4 on Pith
- [29] Jiaming Wang. What stops recursive self-improvement in robotics? Lessons from 123 rounds of agentic skill discov- ery.arXiv preprint arXiv:2609.31760, 2026. 2, 3 on PithPith's reviewarXiv
- [30] Kailin Wang, Haoxiang Jie, Yaoyuan Yan, Zhiyou Heng, and Zhaosong Li. RegenHarness: A robot agent harness with evidence-gated recursive self-improvement.arXiv preprint arXiv:2609.27612, 2026. 2, 3 on PithPith's reviewarXiv
- [31] Peidong Wang, Z. Ma, Y . Chang, X. Luo, X. Yang, S. Feng, Y . Yang, and D. Li. Self-evolving embodied agents via skill- harness evolution.arXiv preprint arXiv:2608.11350, 2026. 2, 3 on PithPith's reviewarXiv
- [32] Xin Wang, Wenhao Wu, Menghao Zhang, Zhi Wang, Kun Shao, Jian Luan, Yang Li, Qing Li, Shangding Gu, Huichi Zhou, Shuqing Shi, Fei Ni, Shuo Lu, Weicheng Meng, Kang Li, Jin Wu, Kang Zhao, Shangmin Guo, Gen Li, Yongqiang Tang, Zhizhong Zhang, Yuan Xie, and Heng Qu. Harnes- sPAI: A... on PithPith's reviewarXiv
- [33] Yen-Jen Wang, Haozhe Jiang, Shuying Deng, Haoru Xue, Weirui Ye, Rocky Duan, Nika Haghtalab, S. Shankar Sas- try, Pieter Abbeel, and Haozhi Qi. Reconstruct, practice, go real: Guided self-improvement for embodied agents.arXiv preprint arXiv:2610.02204, 2026. 2, 3 on PithPith's reviewarXiv
- [34] Wenli Xiao, Jia Xie, Tonghe Zhang, Haotian Lin, Letian Fu, Haoru Xue, Jalen Lu, Yi Yang, Cunxi Dai, Zi Wang, Jimmy Wu, Guanzhi Wang, S. Shankar Sastry, Ken Goldberg, Linxi Fan, Yuke Zhu, and Guanya Shi. ENPIRE: Agentic robot policy self-improvement in the real world.arXiv prep... on PithPith's reviewarXiv
- [35] Hang Yu. You don’t need to stay in the loop: An agentic robotics loop for robot-policy improvement.arXiv preprint arXiv:2608.07555, 2026. 3 on PithPith's reviewarXiv
- [36] Hangfan Zhang, Shao Zhang, Kangcong Li, Chen Zhang, Yang Chen, Yiqun Zhang, Lei Bai, and Shuyue Hu. Self- harness: Harnesses that improve themselves.arXiv preprint arXiv:2606.09498, 2026. 3 on PithPith's reviewarXiv
- [37] Jenny Zhang, Shengran Hu, Cong Lu, Robert Lange, and Jeff Clune. Darwin Godel machine: Open-ended evolution of self-improving agents.arXiv preprint arXiv:2505.22954, on PithPith's reviewarXiv
- [38] Yixian Zhang, Huanming Zhang, Feng Gao, Xiao Li, Zhihao Liu, Yi Nie, Chunyang Zhu, Jiaxing Qiu, Yuchen Yan, Jiyuan Liu, Wenhao Tang, Jiaji Rao, Zhengru Fang, Changxu Wei, Yu Wang, Wenbo Ding, and Chao Yu. Harness VLA: Steer- ing frozen VLAs into reliable manipulation primitive... on PithPith's reviewarXiv
- [39] Xueyang Zhou, Yangming Xu, Guiyao Tie, Yongchao Chen, Guowen Zhang, Duanfeng Chu, Pan Zhou, and Lichao Sun. LIBERO-PRO: Towards robust and fair evaluation of vision-language-action models beyond memorization.arXiv preprint arXiv:2510.03827, 2025. 3 13 on PithPith's reviewarXiv
Cited by 6 Pith papers
Reviewed papers on Pith that cite this one. The full citation record
- Flex-RSI: Benchmarking Recursive Self-Improvement of Frontier Agents cs.AI · Nov 2026
- What Can a VLA Learn from the Agent That Calls It? Task-Level Experience Distillation for Embodied Navigation cs.RO · Nov 2026
- WiSI: Wireless-in-the-Loop Self-Improvement eess.SY · Nov 2026
- Pretraining Knows What, Simulation Teaches How: Compositional Generalization in a Sim-Trained Robot Agent cs.RO · Oct 2026
- OmniBrain-0: A Unified Agentic Model for Embodied Intelligence cs.CL · Oct 2026
- Toward Recursive Self-Improvement in Physical AI cs.RO · Oct 2026
Challenges and discussion
Disagree with the review, or know something it missed? Leave a comment, or challenge this review formally: a challenge becomes a permanent entry in this paper's record, and its decision is published here.
Come back by email
A short list of the week's newly reviewed papers. Nothing is sent until you confirm from your inbox, and every email has a link that stops it. No account.
No comments yet.
Sign in to comment