pith. sign in

arxiv: 2604.07799 · v2 · pith:TYFQR5V6new · submitted 2026-04-09 · 💻 cs.RO · cs.AI

Learning Without Losing Identity: Capability Evolution for Embodied Agents

Xue Qin , Simin Luan , John See , Cong Yang , Zhijun Li This is my paper

Pith reviewed 2026-05-22 10:41 UTC · model grok-4.3

classification 💻 cs.RO cs.AI
keywords embodied agentscapability evolutionagent identitymodular learningsafety constraintsroboticspersistent systems
0
0 comments X

The pith

Embodied agents evolve capabilities separately from their fixed identity to improve over time without instability or safety loss.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper argues that robots should keep a stable cognitive identity while their skills improve through independent modular updates rather than by changing the agent itself. This separation uses versioned Embodied Capability Modules that are learned and refined in a loop of task runs, data collection, and model updates. A safety layer runs on top to block any drift or violations during the process. If the separation works, agents could keep gaining abilities in changing physical settings without periodic resets or identity erosion. Experiments in simulation show success climbing from 32.4 percent to 91.3 percent over twenty iterations while beating baselines and holding policy and safety at zero change.

Core claim

A robot maintains a persistent agent as its cognitive identity while its capabilities evolve through modular Embodied Capability Modules. These modules are learned, refined, and composed via a closed-loop process of task execution, experience collection, model refinement, and module updating, all enforced by a runtime layer that preserves safety and policy constraints.

What carries the argument

Embodied Capability Modules, which are modular and versioned units of embodied functionality that evolve independently while the agent identity stays fixed.

If this is right

  • Task success rates rise from 32.4% to 91.3% across twenty iterations of evolution.
  • The method outperforms both agent-modification approaches and prior skill-learning techniques such as SPiRL and SkiMo.
  • Policy remains unchanged and safety violations stay at zero throughout the process.
  • Decoupling identity from capability evolution supplies a scalable base for long-running embodied systems.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same modular separation could support indefinite operation of service robots in homes without needing identity resets after each new skill is added.
  • It might transfer to other persistent AI agents that must acquire new behaviors while keeping a stable core.
  • Real-world trials in noisy physical settings would test whether the closed-loop updates remain stable outside simulation.

Load-bearing premise

Capabilities can be represented as independent modular units that evolve through task execution and refinement without any interaction that changes the persistent agent identity or breaks safety rules.

What would settle it

Running the framework on a physical robot over many iterations and observing either policy drift or a safety violation during capability updates would disprove the decoupling.

Figures

Figures reproduced from arXiv: 2604.07799 by Cong Yang, John See, Simin Luan, Xue Qin, Zhijun Li.

Figure 1
Figure 1. Figure 1: Capability evolution loop for embodied agents. A persistent agent (blue, top) maintains its identity and decision-making role, while capabilities (ECMs) [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Task success rate vs. evolution iteration for all five methods. Capability Evolution (ours, solid blue) shows sustained improvement across 20 iterations, [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
read the original abstract

Embodied agents are expected to operate persistently in dynamic physical environments, continuously acquiring new capabilities over time. Existing approaches to improving agent performance often rely on modifying the agent itself -- through prompt engineering, policy updates, or structural redesign -- leading to instability and loss of identity in long-lived systems. In this work, we propose a capability-centric evolution paradigm for embodied agents. We argue that a robot should maintain a persistent agent as its cognitive identity, while enabling continuous improvement through the evolution of its capabilities. Specifically, we introduce the concept of Embodied Capability Modules (ECMs), which represent modular, versioned units of embodied functionality that can be learned, refined, and composed over time. We present a unified framework in which capability evolution is decoupled from agent identity. Capabilities evolve through a closed-loop process involving task execution, experience collection, model refinement, and module updating, while all executions are governed by a runtime layer that enforces safety and policy constraints. We demonstrate through simulated embodied tasks that capability evolution improves task success rates from 32.4% to 91.3% over 20 iterations, outperforming both agent-modification baselines and established skill-learning methods (SPiRL, SkiMo), while preserving zero policy drift and zero safety violations. Our results suggest that separating agent identity from capability evolution provides a scalable and safe foundation for long-term embodied intelligence.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper proposes a capability-centric evolution paradigm for embodied agents that maintains a persistent cognitive identity while allowing continuous improvement via modular, versioned Embodied Capability Modules (ECMs). Capabilities evolve in a closed-loop process of task execution, experience collection, model refinement, and module updating, decoupled from the agent core and governed by a runtime safety layer. Simulations of embodied tasks report success rates rising from 32.4% to 91.3% over 20 iterations, outperforming agent-modification baselines and skill-learning methods (SPiRL, SkiMo) with zero policy drift and zero safety violations.

Significance. If the reported gains prove robust, the decoupling of identity from capability evolution could provide a scalable foundation for long-lived embodied systems, avoiding instability from direct agent changes. The explicit runtime enforcement yielding zero violations is a concrete strength that merits further exploration in physical settings.

major comments (2)
  1. [Abstract] Abstract and results section: the central claim of success-rate improvement from 32.4% to 91.3% over 20 iterations, with outperformance over SPiRL and SkiMo, is presented without any description of the number of independent trials, standard deviations, statistical tests, or baseline re-implementation details. This information is load-bearing for evaluating whether the empirical evidence supports the superiority assertion.
  2. [§3] §3 (Framework description): the claim that ECMs evolve as independent modular units without interactions that could alter persistent agent identity or violate safety constraints is introduced but lacks a formal argument or invariant showing that the closed-loop process preserves decoupling under all task executions.
minor comments (1)
  1. The notation for versioned ECMs and the runtime constraint layer could be introduced with a small diagram or pseudocode to improve readability of the unified framework.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and insightful comments on our manuscript. We address each major comment point by point below and indicate the changes we will make in the revised version.

read point-by-point responses

  1. Referee: [Abstract] Abstract and results section: the central claim of success-rate improvement from 32.4% to 91.3% over 20 iterations, with outperformance over SPiRL and SkiMo, is presented without any description of the number of independent trials, standard deviations, statistical tests, or baseline re-implementation details. This information is load-bearing for evaluating whether the empirical evidence supports the superiority assertion.

    Authors: We agree that the current version of the manuscript omits these critical experimental details. In the revised manuscript, we will expand both the abstract and the results section to report that all experiments were conducted over 5 independent trials, include standard deviations for the success rates at each iteration, and present statistical significance tests (paired t-tests with p-values) comparing our method against the baselines. We will also add a dedicated paragraph detailing the re-implementation of SPiRL and SkiMo, confirming that they were evaluated in the identical simulated environment and task suite as our approach. These additions will directly address the load-bearing nature of the empirical claims. revision: yes

  2. Referee: [§3] §3 (Framework description): the claim that ECMs evolve as independent modular units without interactions that could alter persistent agent identity or violate safety constraints is introduced but lacks a formal argument or invariant showing that the closed-loop process preserves decoupling under all task executions.

    Authors: We concur that a formal argument would strengthen the framework section. In the revision, we will insert a new subsection in §3 that introduces a formal invariant: the agent core (defined as the persistent policy parameters and long-term memory state) remains unchanged by construction, with all updates restricted to versioned ECMs. We will prove by induction over the closed-loop iterations that the runtime safety layer enforces this separation for arbitrary task executions, ensuring no policy drift or identity alteration. The proof will be supported by a concise mathematical formulation of the decoupling property. revision: yes

Circularity Check

0 steps flagged

No significant circularity in empirical framework

full rationale

The paper defines a conceptual framework of Embodied Capability Modules (ECMs) decoupled from agent identity, with evolution via a closed-loop process of task execution, experience collection, model refinement, and module updating under runtime safety constraints. The central results are direct empirical outcomes from simulated embodied tasks showing success rate improvement from 32.4% to 91.3% over 20 iterations, outperforming baselines like SPiRL and SkiMo with zero policy drift and zero safety violations. These are presented as simulation measurements rather than quantities derived from fitted parameters, self-referential equations, or self-citations that reduce the claim to its inputs by construction. The derivation chain is self-contained against external simulation benchmarks with no load-bearing steps that collapse into definitions or prior self-work.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The central claim rests on the introduction of a new modular construct (ECMs) and the domain assumption that capabilities can be isolated and evolved independently of identity. No free parameters are explicitly fitted in the abstract, and no external evidence for the new modules is provided.

axioms (1)
  • domain assumption Capabilities can be represented as independent modular units that can be learned, refined, and composed without affecting the agent's persistent cognitive identity.
    This premise is stated directly in the abstract when introducing the capability-centric evolution paradigm.
invented entities (1)
  • Embodied Capability Modules (ECMs) no independent evidence
    purpose: Modular, versioned units of embodied functionality that evolve separately from agent identity.
    ECMs are introduced as the core new construct enabling the decoupling; no independent evidence such as a predicted observable outside the simulation is given.

pith-pipeline@v0.9.0 · 5778 in / 1361 out tokens · 34909 ms · 2026-05-22T10:41:07.009199+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 6 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Harnessing Embodied Agents: Runtime Governance for Policy-Constrained Execution

    cs.RO 2026-04 unverdicted novelty 7.0

    A runtime governance framework for embodied agents achieves 96.2% interception of unauthorized actions and 91.4% recovery success in 1000 simulation trials by externalizing policy enforcement.

  2. EmbodiedGovBench: A Benchmark for Governance, Recovery, and Upgrade Safety in Embodied Agent Systems

    cs.RO 2026-04 unverdicted novelty 6.0

    EmbodiedGovBench is a new benchmark framework that measures embodied agent systems on seven governance dimensions including policy adherence, recovery success, and upgrade safety.

  3. Federated Single-Agent Robotics: Multi-Robot Coordination Without Intra-Robot Multi-Agent Fragmentation

    cs.RO 2026-04 unverdicted novelty 5.0

    Multi-robot coordination is achieved by federating single-agent robot runtimes at the fleet level instead of fragmenting each robot into multiple internal agents.

  4. Federated Single-Agent Robotics: Multi-Robot Coordination Without Intra-Robot Multi-Agent Fragmentation

    cs.RO 2026-04 unverdicted novelty 5.0

    FSAR is a fleet coordination architecture that preserves each robot as a single-agent runtime and achieves multi-robot coordination via capability sharing, delegation, and layered recovery instead of internal agent fr...

  5. ECM Contracts: Contract-Aware, Versioned, and Governable Capability Interfaces for Embodied Agents

    cs.SE 2026-04 unverdicted novelty 5.0

    ECM Contracts define a six-dimensional contract model for embodied capability modules that enables static checks for safe composition, installation, and versioned upgrades in robotics systems.

  6. Harnessing Embodied Agents: Runtime Governance for Policy-Constrained Execution

    cs.RO 2026-04 unverdicted novelty 5.0

    A runtime governance framework for embodied agents intercepts 96.2% of unauthorized actions and achieves 91.4% recovery success in 1000 simulation trials while outperforming baselines.

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages · cited by 4 Pith papers · 5 internal anchors

  1. [1]

    V oyager: An open-ended embodied agent with large language models,

    G. Wang, Y . Xie, Y . Jiang, A. Mandlekar, C. Xiao, Y . Zhu, L. Fan, and A. Anandkumar, “V oyager: An open-ended embodied agent with large language models,” inNeurIPS, 2023

    work page 2023

  2. [2]

    Integrated task and motion planning,

    C. R. Garrett, T. Lozano-Pérez, and L. P. Kaelbling, “Integrated task and motion planning,”Annual Review of Control, Robotics, and Autonomous Systems, vol. 4, pp. 265–293, 2021

    work page 2021

  3. [3]

    robosuite: A Modular Simulation Framework and Benchmark for Robot Learning

    Y . Zhu, J. Wong, A. Mandlekar, R. Martín-Martín, A. Joshi, S. Nasiriany, and Y . Zhu, “robosuite: A modular simulation framework and benchmark for robot learning,” inarXiv preprint arXiv:2009.12293, 2020

    work page internal anchor Pith review Pith/arXiv arXiv 2009

  4. [4]

    Accelerating reinforcement learning with learned skill priors,

    K. Pertsch, Y . Lee, and J. J. Lim, “Accelerating reinforcement learning with learned skill priors,” inCoRL, 2021

    work page 2021

  5. [5]

    Skill-based model-based reinforcement learning,

    L. X. Shi, J. J. Lim, and Y . Lee, “Skill-based model-based reinforcement learning,” inCoRL, 2023

    work page 2023

  6. [6]

    A generalist agent,

    S. Reed, K. Zolna, E. Parisotto, S. G. Colmenarejo, A. Novikov,et al., “A generalist agent,” inTMLR, 2022

    work page 2022

  7. [7]

    Between MDPs and semi- MDPs: A framework for temporal abstraction in reinforcement learning,

    R. S. Sutton, D. Precup, and S. Singh, “Between MDPs and semi- MDPs: A framework for temporal abstraction in reinforcement learning,” Artificial Intelligence, vol. 112, no. 1–2, pp. 181–211, 1999

    work page 1999

  8. [8]

    FeUdal networks for hierarchical reinforcement learning,

    A. S. Vezhnevets, S. Osindero, T. Schaul, N. Heess, M. Jaderberg, D. Silver, and K. Kavukcuoglu, “FeUdal networks for hierarchical reinforcement learning,” inICML, 2017

    work page 2017

  9. [9]

    The option-critic architecture,

    P.-L. Bacon, J. Harb, and D. Precup, “The option-critic architecture,” in AAAI, 2017

    work page 2017

  10. [10]

    Data-efficient hierarchical reinforcement learning,

    O. Nachum, S. Gu, H. Lee, and S. Levine, “Data-efficient hierarchical reinforcement learning,” inNeurIPS, 2018

    work page 2018

  11. [11]

    Diversity is all you need: Learning skills without a reward function,

    B. Eysenbach, A. Gupta, J. Ibarz, and S. Levine, “Diversity is all you need: Learning skills without a reward function,” inICLR, 2019

    work page 2019

  12. [12]

    Do as i can, not as i say: Grounding language in robotic affordances,

    M. Ahn, A. Brohan, N. Brown, Y . Chebotar, O. Cortes,et al., “Do as i can, not as i say: Grounding language in robotic affordances,” inCoRL, 2022

    work page 2022

  13. [13]

    Code as policies: Language model programs for embodied control,

    J. Liang, W. Huang, F. Xia, P. Xu, K. Hausman, B. Ichter, P. Florence, and A. Zeng, “Code as policies: Language model programs for embodied control,” inICRA, 2023

    work page 2023

  14. [14]

    Inner monologue: Em- bodied reasoning through planning with language models,

    W. Huang, F. Xia, T. Xiao, H. Chan, J. Liang, P. Florence, A. Zeng, J. Tompson, I. Mordatch, Y . Chebotar,et al., “Inner monologue: Em- bodied reasoning through planning with language models,” inCoRL, 2022

    work page 2022

  15. [15]

    PaLM-E: An embodied multimodal language model,

    D. Driess, F. Xia, M. S. M. Sajjadi, C. Lynch, A. Chowdhery,et al., “PaLM-E: An embodied multimodal language model,” inICML, 2023

    work page 2023

  16. [16]

    RT-1: Robotics transformer for real-world control at scale,

    A. Brohan, N. Brown, J. Carbajal, Y . Chebotar, J. Dabis, C. Finn,et al., “RT-1: Robotics transformer for real-world control at scale,” inRSS, 2023

    work page 2023

  17. [17]

    RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control

    A. Brohan, N. Brown, J. Carbajal, Y . Chebotar, X. Chen,et al., “RT- 2: Vision-language-action models transfer web knowledge to robotic control,”arXiv preprint arXiv:2307.15818, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  18. [18]

    Learning modular neural network policies for multi-task and multi-robot transfer,

    C. Devin, A. Gupta, T. Darrell, P. Abbeel, and S. Levine, “Learning modular neural network policies for multi-task and multi-robot transfer,” inICRA, 2017

    work page 2017

  19. [19]

    Robocat: A self-improving foundation agent for robotic manipulation,

    K. Bousmalis, O. Vinyals, K. Zidek,et al., “RoboCat: A self- improving generalist agent for robotic manipulation,”arXiv preprint arXiv:2306.11706, 2023

    work page arXiv 2023

  20. [20]

    Lifelong robot library learning: Bootstrap- ping composable and generalizable skills for embodied control with language models,

    G. Tziafas and H. Kasaei, “Lifelong robot library learning: Bootstrap- ping composable and generalizable skills for embodied control with language models,” inICRA, 2024

    work page 2024

  21. [21]

    A survey on the lifecycle of microservices,

    S. Dragicevic and S. Celar, “A survey on the lifecycle of microservices,” IEEE Access, vol. 11, pp. 30497–30510, 2023

    work page 2023

  22. [22]

    Continual lifelong learning with neural networks: A review,

    G. I. Parisi, R. Kemker, J. L. Part, C. Kanan, and S. Wermter, “Continual lifelong learning with neural networks: A review,”Neural Networks, vol. 113, pp. 54–71, 2019

    work page 2019

  23. [23]

    Learning without forgetting,

    Z. Li and D. Hoiem, “Learning without forgetting,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 40, no. 12, pp. 2935– 2947, 2017

    work page 2017

  24. [24]

    Lifelong robot learning,

    S. Thrun and T. M. Mitchell, “Lifelong robot learning,”Robotics and Autonomous Systems, vol. 15, no. 1–2, pp. 25–46, 1995

    work page 1995

  25. [25]

    Three types of incremental learning,

    G. M. van de Ven, T. Tuytelaars, and A. S. Tolias, “Three types of incremental learning,”Nature Machine Intelligence, vol. 4, pp. 1185– 1197, 2022

    work page 2022

  26. [26]

    Overcoming catastrophic forgetting in neural networks,

    J. Kirkpatrick, R. Pascanu, N. Rabinowitz, J. Veness, G. Desjardins, A. A. Rusu, K. Milan, J. Quan, T. Ramalho, A. Grabska-Barwinska, et al., “Overcoming catastrophic forgetting in neural networks,”Proceed- ings of the National Academy of Sciences, vol. 114, no. 13, pp. 3521– 3526, 2017

    work page 2017

  27. [27]

    Continual learning through synaptic intelligence,

    F. Zenke, B. Poole, and S. Ganguli, “Continual learning through synaptic intelligence,” inICML, 2017

    work page 2017

  28. [28]

    Gradient episodic memory for continual learning,

    D. Lopez-Paz and M. Ranzato, “Gradient episodic memory for continual learning,” inNeurIPS, 2017

    work page 2017

  29. [29]

    Progressive Neural Networks

    A. A. Rusu, N. C. Rabinowitz, G. Desjardins, H. Soyer, J. Kirkpatrick, K. Kavukcuoglu, R. Pascanu, and R. Hadsell, “Progressive neural networks,” inarXiv preprint arXiv:1606.04671, 2016

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  30. [30]

    PathNet: Evolution Channels Gradient Descent in Super Neural Networks

    C. Fernando, D. Banarse, C. Blundell, Y . Zwols, D. Ha, A. A. Rusu, A. Pritzel, and D. Wierstra, “PathNet: Evolution channels gradient descent in super neural networks,”arXiv preprint arXiv:1701.08734, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  31. [31]

    Continual learning for robotics: Definition, framework, learning strategies, opportunities and challenges,

    T. Lesort, V . Lomonaco, A. Stoian, D. Maltoni, D. Filliat, and N. Díaz- Rodríguez, “Continual learning for robotics: Definition, framework, learning strategies, opportunities and challenges,”Information Fusion, vol. 58, pp. 52–68, 2020

    work page 2020

  32. [32]

    Continual world: A robotic benchmark for continual reinforcement learning,

    M. Wołczyk, M. Zaj ˛ ac, R. Danielczuk,et al., “Continual world: A robotic benchmark for continual reinforcement learning,” inNeurIPS, 2021

    work page 2021

  33. [33]

    ReAct: Synergizing reasoning and acting in language models,

    S. Yao, J. Zhao, D. Yu, N. Du, I. Shafran, K. Narasimhan, and Y . Cao, “ReAct: Synergizing reasoning and acting in language models,” inICLR, 2023

    work page 2023

  34. [34]

    Language models as zero-shot planners: Extracting actionable knowledge for embodied agents,

    W. Huang, P. Abbeel, D. Pathak, and I. Mordatch, “Language models as zero-shot planners: Extracting actionable knowledge for embodied agents,” inICML, 2022

    work page 2022

  35. [35]

    Toolformer: Language models can teach themselves to use tools,

    T. Schick, J. Dwivedi-Yu, R. Dessì, R. Raileanu, M. Lomeli,et al., “Toolformer: Language models can teach themselves to use tools,” in NeurIPS, 2023

    work page 2023

  36. [36]

    DEPS: Describe, explain, plan and select: Interactive planning with large language models enables open-world multi-task agents,

    Z. Wang, S. Cai, G. Chen, A. Liu, X. Ma, and Y . Liang, “DEPS: Describe, explain, plan and select: Interactive planning with large language models enables open-world multi-task agents,” inNeurIPS, 2023

    work page 2023

  37. [37]

    Reflexion: Language agents with verbal reinforcement learning,

    N. Shinn, F. Cassano, A. Gopinath, K. Narasimhan, and S. Yao, “Reflexion: Language agents with verbal reinforcement learning,” in NeurIPS, 2023

    work page 2023

  38. [38]

    A comprehensive survey on safe reinforce- ment learning,

    J. García and F. Fernández, “A comprehensive survey on safe reinforce- ment learning,”Journal of Machine Learning Research, vol. 16, no. 42, pp. 1437–1480, 2015

    work page 2015

  39. [39]

    Safe learning in robotics: From learning-based control to safe reinforcement learning,

    L. Brunke, M. Greeff, A. W. Hall, Z. Yuan, S. Zhou, J. Panerati, and A. P. Schoellig, “Safe learning in robotics: From learning-based control to safe reinforcement learning,”Annual Review of Control, Robotics, and Autonomous Systems, vol. 5, pp. 411–444, 2022

    work page 2022

  40. [40]

    Constrained policy optimization,

    J. Achiam, D. Held, A. Tamar, and P. Abbeel, “Constrained policy optimization,” inICML, 2017

    work page 2017

  41. [41]

    Safe reinforcement learning via shielding,

    M. Alshiekh, R. Bloem, R. Ehlers, B. Könighofer, S. Niekum, and U. Topcu, “Safe reinforcement learning via shielding,” inAAAI, 2018

    work page 2018

  42. [42]

    Safety-gymnasium: A unified safe reinforcement learning benchmark,

    J. Ji, B. Zhang, J. Zhou, J. Pan,et al., “Safety-gymnasium: A unified safe reinforcement learning benchmark,” inNeurIPS Datasets and Benchmarks Track, 2024

    work page 2024

  43. [43]

    AEROS: A Single-Agent Operating Architecture with Embodied Capability Modules

    X. Qin, S. Luan, C. Yang, and Z. Li, “AEROS: Agent execu- tion runtime operating system for embodied robots,”arXiv preprint arXiv:2604.07039, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026