Pith. sign in

REVIEW 4 major objections 6 minor 29 references

This paper argues that when a robot's action proposal is rejected by a safety tester, the correct response is causal diagnosis — identifying which parameter caused the failure and what corrective value would maximize the probability of pass

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

A causal circuit built from a Joint Probability Tree lets a robot correct rejected motion plans in one shot, cutting failed safety-test attempts by 10–37% in simulation.

T0 review reviewed 2026-08-02 challenge →

load-bearing objection The closed-loop recovery system is a plausible density-guided heuristic; the causal diagnosis claim fails against the paper's own equations. the 4 major comments →

arxiv 2607.14826 v1 pith:IRPUYUQ7 submitted 2026-07-16 cs.RO cs.AI

Interventional Causal Circuits for Safe Robot Action Testing and Failure Recovery

Naren Vasantakumaar , Tom Schierenbeck , Michael Beetz This is my paper
classification cs.RO cs.AI MSC 68T4062D20
keywords causal inferenceprobabilistic circuitsjoint probability treesrobot safetyfailure recoveryinterventional distributionsbackdoor adjustmentsafe physical AI
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The reading

The paper aims to replace blind resampling in robot planning-testing loops with something more informative: when a proposed action fails a safety test, diagnose which action parameter caused the failure and which corrected value makes passing most likely. The central construction is a Causal Circuit, obtained from a Joint Probability Tree (the fitted planning distribution) by imposing a variable-tree structure and verifying a support-disjointness condition, so that interventional queries of the form 'what if this parameter were forced to that value' can be answered exactly in polynomial time. This turns each rejection into a one-shot correction: the primary cause variable is constrained to a recommended interval and the other parameters are resampled from the conditional distribution. The paper reports that in a simulation of a pick-and-place task this reduces failed attempts by 10.3% under a high-quality planning distribution and by 37% under a degraded one, while keeping worst-case attempts per iteration low and producing interpretable causal reports. A sympathetic reader would care because this offers a principled, data-efficient way to recover from failures without retraining or a separately trained failure model.

Core claim

The central claim is that a Joint Probability Tree fitted to successful executions can be extended without retraining into a Causal Circuit by imposing a Marginal-Deterministic Variable Tree and verifying that every sum unit partitions its children's support disjointly on each declared cause variable. When this support-determinism condition holds, the backdoor adjustment formula reduces to a weighted sum over non-overlapping leaf regions, so the interventional probability P(Y | do(x_i = v)) is computed exactly and in polynomial time. At runtime, a rejected plan is diagnosed by evaluating for each cause variable the interventional probability mass in a narrow interval around its observed valu

What carries the argument

The load-bearing object is the Causal Circuit: a Joint Probability Tree restructured so that its sum units partition the support of each designated cause variable disjointly (called support determinism), with a cause-priority order set by normalized average treatment effect. This structural property makes the backdoor adjustment sum collapse to a sum over non-overlapping leaf regions, enabling exact, polynomial-time computation of interventional probabilities P(Y | do(x_i = v)) for every cause variable. The circuit is verified before deployment, providing a pre-runtime certificate that all interventional queries are tractable. At failure time, the circuit evaluates a diagnostic score rho_i(x

Load-bearing premise

The entire causal diagnosis rests on the unstated assumption that the hand-designated cause variables and effect satisfy the backdoor criterion, meaning there is no unobserved confounder influencing both a cause parameter and the success outcome; if that fails, the computed 'interventional' probabilities are not identifiable from the fitted distribution and the correction may be steered by correlation rather than causation.

What would settle it

Run the pipeline on a task with a deliberately hidden confounder (for example, object weight affecting both approach position and success). If a randomized intervention that forces the recommended value gives a success rate far below the circuit's predicted interventional probability, the backdoor adjustment is invalid. Alternatively, compare the circuit's correction against a ground-truth interventional estimator computed by actually forcing cause variables to values in a simulator.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If a Joint Probability Tree satisfies the support-determinism condition, every failed action proposal can be turned into a targeted corrective attempt without retraining, additional data collection, or extra simulation rollouts.
  • The pre-deployment verification step means the system can guarantee before operation that all interventional queries — and hence all failure diagnoses — are computable exactly and in polynomial time.
  • Out-of-support proposals, where the observed parameters lie entirely outside the training leaves, are automatically detected and excluded from correction, so the system never extrapolates beyond observed data.
  • In the simulated pick-and-place task, the causal circuit reduces total failed attempts by 10.3% under a dense planning distribution and by 37% under a degraded distribution, and cuts the worst-case attempts per iteration from 10 to 3 in the degraded case.
  • Each rejected plan yields a structured, interpretable report naming the primary cause variable, its observed value, and the recommended corrective region, supporting operator oversight without a separately trained failure model.

Where Pith is reading between the lines

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

  • If unobserved confounders exist between the action parameters and the success outcome, the backdoor adjustment used here is not identifiable from the observational distribution alone; the 'causal' diagnosis would silently reduce to ordinary conditioning, so the safety certificate would not license a causal reading of the numbers.
  • The same construction should transfer to any continuous-parameter action space, so a testable extension is to run the pipeline on tasks with non-positional parameters (gripper force, object weight, contact dynamics) where the paper's current single pick-and-place evaluation does not yet apply.
  • A natural follow-up is to make the correction chaining with a decision cost: since the paper deliberately limits to one-shot corrections to avoid committing to a bad recommendation, a variant could weigh the expected cost of another tester call against the predicted success gain from a second correction.
  • Because the interventional distribution is computed exactly, the diagnostic score rho_i could be reused for active learning, selecting which parameter to vary next in an experiment, turning the failure-recovery loop into a data-efficiency tool.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. This paper proposes a closed-loop hypothesis-test-debug architecture for robot action planning: a Joint Probability Tree (JPT) fitted to successful executions samples candidate action parameters, a tester rejects unsafe candidates, and on rejection an 'Interventional Causal Circuit' — the JPT augmented with a Marginal-Deterministic Vtree (MdVtree) over an ATE-norm-ranked cause set — is claimed to compute exact polytime interventional queries, identify the primary cause variable, and recommend a corrective region for a one-shot re-attempt. The authors report ROS2 pick-and-place experiments in which the circuit reduces total failed attempts by 10.3% with a high-quality JPT and by 37% with a degraded JPT relative to blind resampling, and produces structured, interpretable causal failure reports. The central claim is that rejection triggers causal diagnosis under the interventional distribution rather than a naive resample.

Significance. If the causal claims held, this would be a noteworthy contribution: runtime, retraining-free causal diagnosis with pre-deployment tractability certificates, interpretable failure reports, and automatic out-of-support exclusion are all attractive properties for safety-critical planning-testing loops, and the combination of JPTs with Wang and Kwiatkowska's MdVtree circuits [1] is a sensible and potentially useful engineering idea. The paper is also commendable for shipping an experimental setup and promising code, and the one-shot, non-chaining correction design prudently limits the risk of committing to a bad recommendation. However, the central causal semantics are not supported by the manuscript's own equations and inputs: the implemented diagnosis reduces to the observational marginal of the fitted JPT, the backdoor adjustment is never justified, and the training-on-success data make the success probability degenerate within support. The measured retry reductions therefore do not validate causal diagnosis, and the paper's contribution as stated is not established.

major comments (4)
  1. [Sec. IV-D, Eq. (3)] The primary-cause statistic is defined as the mass of the 'marginal interventional distribution' of x_i around the observed value, ρ_i(x̂_i) = ∫ p^{do(x_i)}(x_i) dx_i (Eq. 3). Under the paper's own construction of the interventional circuit (Eq. 2) — a SumUnit over cause-support regions R_k with weights P(x_i ∈ R_k), each branch being ProductUnit(p(x_i | x_i ∈ R_k), p(Y | x_i ∈ R_k)) — marginalizing onto x_i and Y yields exactly the observational marginal p(x_i): the effect factors integrate to unity and the mixture recombines to the original density. Thus ρ_i is a local density check of the fitted JPT, not a causal effect of x_i on Y. If do(x_i = v) is read literally as a hard intervention, p^{do(x_i)} is a point mass and ρ_i is 0 or 1, incapable of ranking causes or recommending corrections. Neither reading supports the claimed causal diagnosis.
  2. [Sec. III vs. Sec. IV-D] The paper motivates diagnosis with π_i(v) = P(Y ≥ τ | do(x_i = v), C_{-i} = x̂_{-i}) (Sec. III): the primary cause should minimize π_i(x̂_i) and the correction should maximize π_i(v). The implemented diagnosis (Sec. IV-D) never computes π_i; it computes only ρ_i (Eq. 3) and the recommended region R* = argmax_{R_k} P(x_i ∈ R_k | do(x_i)), neither of which conditions on the effect Y. R* maximizes the marginal probability mass of x_i, not the probability of passing the tester. The 'causal report' therefore contains no information about the outcome variable, and the stated objective is never evaluated.
  3. [Sec. IV-C, Eq. (1)] The backdoor adjustment is asserted without identifying a valid adjustment set Z or arguing the backdoor criterion. The only inputs are successful executions plus a hand designation of causes and effect (Sec. III); no causal graph is given, and support determinism verification (Sec. IV-B) certifies only disjointness of leaf supports — a computational property — not causal sufficiency or absence of unobserved confounders. Without confounders Eq. (1) collapses to observational conditioning P(Y ≥ τ | x_i = v, Z = z); with confounders it is invalid without a justified adjustment set. Moreover, fitting the JPT exclusively to successful executions (Sec. III) means every training realization has Y ≥ τ, so within the training support the success probability is 1 by construction and the effect branch of the circuit carries no failure signal; the 'interventional success probability' is degenerate.
  4. [Sec. V, Tables I and II] The experiments do not test the causal claim. The tester output enters only as a binary trigger of the diagnosis; the diagnosis itself (Eq. 3, R*) is computed from the JPT alone and never conditions on Y. A non-causal baseline that flags low marginal density p(x_i) and corrects toward the highest-density region would produce identical corrections, yet no such baseline or ablation of the 'causal' components (ATE ordering, MdVtree, backdoor sum) is reported, and there is no verification that the named primary cause is actually the parameter responsible for the simulator rejection. The 10.3% and 37% retry reductions (Tables I and II) therefore support at most a density-guided correction heuristic, not causal diagnosis; no statistical significance or error bars are reported for either comparison.
minor comments (6)
  1. [Tables I and II] Metric definitions are unclear — e.g., 'Failed iterations 0' with 95 'Corrected attempts' in Table I; specify whether the initial rejected plan is counted in 'attempts per recovery' and how 'corrected attempts' relates to 'failed attempts.'
  2. [Footnote 1] The GitHub URL ('https://github.com/Narenvasant/causal reasoning') contains a space and appears broken; a working link is needed given the paper's reproducibility claim.
  3. [Sec. II and Sec. VI] 'Our work experiments this theoretical framework' is informal — consider 'applies' or 'evaluates.' The Limitations section is a single sentence with a grammatical error ('transfer the experimentation to a real world...') and is too thin for a safety-critical framework paper.
  4. [Eq. (1) and Fig. 2] The adjustment set Z is never explicitly defined and is conflated with the remaining cause variables C_{-i} (Fig. 2 uses Z for 'the remaining cause variables'); unify the notation and state the intended causal graph before presenting Eq. (1).
  5. [Sec. III / Sec. IV-A] The ATE_norm ordering is never defined: no estimator, normalization, or adjustment is specified, so the construction of the MdVtree leaves is not reproducible. The threshold τ on the continuous success proxy Y is also never given.
  6. [Fig. 2] The figure depicts do(x_0) as removing incoming edges, but the described construction (Eq. 2) does not perform edge removal; it re-weights and truncates leaf regions. The figure's causal semantics are not realized by the implementation.

Circularity Check

1 steps flagged

The 'interventional' diagnosis is an observational density check: Eq. (3)'s ρ_i equals the JPT marginal p(x_i) under Eq. (2), so cause ranking and correction reduce to fitted inputs rather than causal effects.

specific steps
  1. fitted input called prediction [Sec. IV-C Eq. (2) and Sec. IV-D Eq. (3)]
    "ρ_i(x̂_i) = ∫_{x̂_i−ε}^{x̂_i+ε} p^{do(x_i)} dx_i ... The primary cause is the variable with the lowest ρ_i(x̂_i) ... SumUnit(P(x_i∈R_k)) → ProductUnit(p(x_i|x_i∈R_k), p(Y|x_i∈R_k))"

    By Eq. (2), the 'interventional circuit' is a mixture over cause-support regions R_k with weights P(x_i∈R_k) and branch densities p(x_i|x_i∈R_k). Summing the cause branch over k returns exactly the JPT's observational marginal p(x_i). Hence p^{do(x_i)} in Eq. (3) is p(x_i), and ρ_i measures local density under the fitted training distribution, not any effect of x_i on Y. The recommended region R* = argmax P(x_i∈R_k|do(x_i)) likewise selects the region with largest observational mass. Y never appears in ρ_i or R*, so the 'primary cause' and 'corrective region' are functions of the fitted JPT alone; the interventional claim reduces to a relabeling of the observational marginal.

full rationale

The empirical results (10.3% and 37% reductions in failed attempts) are real experimental comparisons and are not themselves circular: they measure a closed-loop policy against blind resampling. However, the paper's central causal-diagnosis claim is not computed as described. The only implemented diagnosis is Eq. (3), and Eq. (2) makes the 'marginal interventional distribution' equal to the trained JPT's observational marginal by construction. Thus the ranking of causes and the recommended correction reduce to fitted inputs relabeled as interventional queries. The paper also defines π_i(v)=P(Y≥τ|do(x_i=v),C_{-i}=x̂_{-i}) in Sec. III but never evaluates this quantity in Sec. IV-D; the backdoor adjustment in Eq. (1) is asserted without a justified causal graph, and with no confounders it collapses to observational conditioning. There is no load-bearing self-citation chain: the theoretical basis [1] is external work. The score is 7 rather than higher because the measured efficiency gains have independent content as a density-guided resampling heuristic; what reduces by construction is the causal interpretation and diagnosis mechanism.

Axiom & Free-Parameter Ledger

4 free parameters · 5 axioms · 1 invented entities

The central claim rests on: a hand-designated cause/effect order without modeled confounders; Wang & Kwiatkowska's tractability theorems taken as background; the empirically-passing Q-determinism verification (no fallback defined if it fails); unvalidated sim-to-sim transfer of the fitted JPT via coordinate remapping; and the tester's labels as safety ground truth. Free parameters include the ATE-based causal ordering, JPT hyperparameters (δ also sets the correction step), and unreported ε and τ. No new physical entities are postulated; the Interventional Causal Circuit is a software construction whose hard-intervention semantics are not independently evidenced.

free parameters (4)
  • ATE_norm causal priority ordering (C order) = counter_x > table_x > arm > counter_y > table_y
    Computed from normalised average treatment effect on the training set (Sec. IV-A); this ordering defines the MdVtree of the Causal Circuit, so the entire causal structure is a data-derived input.
  • JPT hyperparameters δ, n_min, σ = δ=0.005 n_min=25 (high-quality); δ=0.15 n_min=600 σ=0.18 (degraded)
    Hand-chosen experimental settings (Sec. V); δ also sets the one-shot correction interval width, so a modelling choice directly controls the correction step.
  • Query resolution ε in Eq. (3) = unspecified
    ρ_i — the measure that picks the primary cause — depends on ε, stated only as 'matched to the JPT leaf precision' (Sec. IV-D); no value or sensitivity analysis is given.
  • Success threshold τ on effect Y = unspecified
    The interventional probability P(Y≥τ|do(...)) and every reported success/failure number depend on τ (the milk placing-height proxy); the value is never reported (Sec. III, V).
axioms (5)
  • domain assumption The designated causes C and effect Y form a causally sufficient model (no unobserved confounders) and the backdoor criterion holds for Z = C\{x_i}.
    Sec. III states the only task-specific inputs are successful executions and a designation of causes/effect; no confounders are modeled and no backdoor-path argument is given. Without this assumption Eq. (1) is not identified; with it, the formula reduces to observational conditioning.
  • standard math MdVtree-imposed marginal determinism yields exact polytime interventional inference via backdoor adjustment (theorems of Wang & Kwiatkowska [1]).
    The paper's tractability claims (Sec. IV-B/C) are inherited from the AISTATS 2023 reference [1]; accepted as background theory, not re-derived here.
  • domain assumption The fitted JPT satisfies marginal (Q-)determinism w.r.t. the ATE-ranked MdVtree, and the pre-deployment verification passes.
    Sec. IV-B admits determinism only 'tends to hold' and reports it passed experimentally; no repair or fallback is defined if verification fails, so the construction's validity is contingent on an empirical property of one dataset.
  • domain assumption A JPT trained on 1,742 successful executions in the open world remains a valid planning distribution in the apartment world after coordinate remapping.
    Sec. V transfers without retraining; no distribution-shift analysis or coverage validation is given, yet all results depend on the remapped model's fidelity.
  • domain assumption Tester pass/fail labels in the deployment world are an accurate safety ground truth.
    The framework never questions the tester (Sec. III); it optimizes probability of passing this tester, which is not necessarily physical safety.
invented entities (1)
  • Interventional Causal Circuit (region-truncated circuit of Eq. (2)) no independent evidence
    purpose: Compute P(Y|do(x_i)) queries and the corrective region R* used for one-shot correction.
    Introduced by this paper as a construction over the fitted JPT; its interpretation is ambiguous (hard intervention gives a degenerate x_i marginal, making Eq. (3) ill-defined, while Eq. (2) matches a region-truncated observational object) and no external benchmark validates its causal semantics beyond the paper's own simulation.

reviewed 2026-08-02 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Interventional Causal Circuits for Safe Robot Action Testing and Failure Recovery." pith.science (2026). https://pith.science/paper/IRPUYUQ7

@misc{pith2026260714826,
  author       = {Pith},
  title        = {Pith review of: Interventional Causal Circuits for Safe Robot Action Testing and Failure Recovery},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IRPUYUQ7}},
  note         = {Machine review of arXiv:2607.14826}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Safe physical AI for robot actions are required not only likely to succeed but tested to be safe before execution. In practice, however, formal testing of motion parameters is computationally expensive, and the cost scales poorly with the dimensionality of the action space. When a proposed action is rejected by a tester, the naive response is to resample blindly until a passing candidate is found. This is wasteful, uninformative, and offers no convergence. We argue that rejection should instead trigger causal diagnosis: a principled identification of which action parameter caused the failure and what corrective value maximises the probability of passing testing under the interventional probability distribution. We propose a closed-loop framework that couples a Joint Probability Tree (JPT) with a Causal Circuit derived from a Marginal-Deterministic Variable Tree, enabling exact polytime computation without retraining, or additional data collection. The framework validates tractability of all interventional queries before the robot begins operating, and out-of-support candidates are detected and excluded from correction automatically. We perform experiments in a ROS2 simulation environment, and the framework demonstrates complementary roles across quality of distribution: under a high-quality JPT, the Causal Circuit reduces failed attempts by 10.3% and under a degraded JPT, it reduces total failed attempts by 37%. Every rejected plan produces a structured, interpretable causal report naming the primary cause variable, its observed value, and the recommended corrective region, supporting operator oversight and autonomous recovery without a separately trained failure model.

Figures

Figures reproduced from arXiv: 2607.14826 by Michael Beetz, Naren Vasantakumaar, Tom Schierenbeck.

Figure 1
Figure 1. Figure 1: The figure represents the proposed closed-loop framework using causal [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Construction of the Interventional Causal Circuit. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

29 extracted references · 6 linked inside Pith

  1. [1]

    Compositional probabilistic and causal inference using tractable circuit models,

    B. Wang and M. Kwiatkowska, “Compositional probabilistic and causal inference using tractable circuit models,” inProc. 26th Int. Conf. Artif. Intell. Stat. (AISTATS), vol. 206, 2023, pp. 9488–9498

    2023

  2. [2]

    Pearl,Causality: Models, Reasoning, and Inference, 2nd ed

    J. Pearl,Causality: Models, Reasoning, and Inference, 2nd ed. Cam- bridge Univ. Press, 2009

    2009

  3. [3]

    Causal inference in statistics: An overview,

    J. Pearl, “Causal inference in statistics: An overview,”Stat. Surv., vol. 3, pp. 96–146, 2009

    2009

  4. [4]

    Probabilistic circuits: A unifying framework for tractable probabilistic models,

    Y . Choi, A. Vergari, and G. Van den Broeck, “Probabilistic circuits: A unifying framework for tractable probabilistic models,” Tech. Rep., UCLA, 2020

    2020

  5. [5]

    Joint probability trees,

    M. Nygaet al., “Joint probability trees,”arXiv:2302.07167, 2023

    Pith/arXiv arXiv 2023

  6. [6]

    Concrete problems in AI safety,

    D. Amodeiet al., “Concrete problems in AI safety,” arXiv:1606.06565, 2016

    Pith/arXiv arXiv 2016

  7. [7]

    Recover: A neuro-symbolic framework for failure detection and recovery,

    A. Parasharet al., “Recover: A neuro-symbolic framework for failure detection and recovery,”arXiv:2404.00756, 2024

    Pith/arXiv arXiv 2024

  8. [8]

    Diagnose, correct, and learn from manipulation failures via visual symbols,

    Z. Liet al., “Diagnose, correct, and learn from manipulation failures via visual symbols,”arXiv:2512.02787, 2025

    arXiv 2025

  9. [9]

    Learning to recover from plan execution errors during robot manipulation: A neuro-symbolic approach,

    H. Sunget al., “Learning to recover from plan execution errors during robot manipulation: A neuro-symbolic approach,”arXiv:2405.18948, 2024

    Pith/arXiv arXiv 2024

  10. [10]

    Learning sampling distribu- tions for robot motion planning,

    B. Ichter, J. Harrison, and M. Pavone, “Learning sampling distribu- tions for robot motion planning,” inProc. IEEE Int. Conf. Robot. Autom. (ICRA), 2018, pp. 7087–7094

    2018

  11. [11]

    Hierarchical rejection sampling for informed kinodynamic planning in high-dimensional spaces,

    T. Kunz, A. Thomaz, and H. Christensen, “Hierarchical rejection sampling for informed kinodynamic planning in high-dimensional spaces,” inProc. IEEE Int. Conf. Robot. Autom. (ICRA), 2016, pp. 89– 96

    2016

  12. [12]

    Synthesis for robots: Guarantees and feedback for robot behavior,

    H. Kress-Gazit, M. Lahijanian, and V . Raman, “Synthesis for robots: Guarantees and feedback for robot behavior,”Annu. Rev. Control Robot. Auton. Syst., vol. 1, pp. 211–236, 2018

    2018

  13. [13]

    Sampling-based motion planning: A comparative review,

    A. Orthey, C. Chamzas, and L. E. Kavraki, “Sampling-based motion planning: A comparative review,”Annu. Rev. Control Robot. Auton. Syst., 2024

    2024

  14. [14]

    Learning feasibility for task and motion planning in tabletop environments,

    R. Chitniset al., “Learning feasibility for task and motion planning in tabletop environments,”IEEE Robot. Autom. Lett., vol. 4, no. 2, pp. 1255–1262, 2019

    2019

  15. [15]

    Motion planning around obstacles with convex optimization,

    T. Marcucciet al., “Motion planning around obstacles with convex optimization,”Sci. Robot., vol. 8, no. 84, 2023

    2023

  16. [16]

    Integrated task and motion planning,

    C. R. Garrettet al., “Integrated task and motion planning,”Annu. Rev. Control Robot. Auton. Syst., vol. 4, pp. 265–293, 2021

    2021

  17. [17]

    Robot manipulation in everyday activities with the CRAM 2.0 cognitive architecture and generalized action plans,

    M. Beetz, G. Kazhoyan, and D. Vernon, “Robot manipulation in everyday activities with the CRAM 2.0 cognitive architecture and generalized action plans,”Cogn. Syst. Res., vol. 92, p. 101375, 2025

    2025

  18. [18]

    Toward causal representation learning,

    B. Sch ¨olkopfet al., “Toward causal representation learning,”Proc. IEEE, vol. 109, no. 5, pp. 612–634, 2021

    2021

  19. [19]

    CausalWorld: A robotic manipulation benchmark for causal structure and transfer learning,

    O. Ahmedet al., “CausalWorld: A robotic manipulation benchmark for causal structure and transfer learning,” inProc. Int. Conf. Learn. Represent. (ICLR), 2021

    2021

  20. [20]

    Causal reasoning in simulation for structure and transfer learning of robot manipulation policies,

    T. E. Leeet al., “Causal reasoning in simulation for structure and transfer learning of robot manipulation policies,”IEEE Robot. Autom. Lett., vol. 7, no. 2, pp. 3840–3847, 2022

    2022

  21. [21]

    Interventional sum-product networks: Causal inference with tractable probabilistic models,

    M. Ze ˇcevi´cet al., “Interventional sum-product networks: Causal inference with tractable probabilistic models,” inAdv. Neural Inf. Process. Syst. (NeurIPS), 2021

    2021

  22. [22]

    Peters, D

    J. Peters, D. Janzing, and B. Sch ¨olkopf,Elements of Causal Inference: Foundations and Learning Algorithms. MIT Press, 2017

    2017

  23. [23]

    A unified framework for real-time failure handling in robotics using vision-language models, reactive planner and behavior trees,

    F. Ahmadet al., “A unified framework for real-time failure handling in robotics using vision-language models, reactive planner and behavior trees,”arXiv:2503.15202, 2025

    Pith/arXiv arXiv 2025

  24. [24]

    Execution monitoring in robotics: A survey,

    O. Pettersson, “Execution monitoring in robotics: A survey,”Robot. Auton. Syst., vol. 53, no. 2, pp. 73–88, 2005

    2005

  25. [25]

    Plan execution monitor- ing through detection of unmet expectations about action outcomes,

    J. P. Mendoza, M. Veloso, and R. Simmons, “Plan execution monitor- ing through detection of unmet expectations about action outcomes,” inProc. IEEE Int. Conf. Robot. Autom. (ICRA), 2015, pp. 3247–3252

    2015

  26. [26]

    Improving robot plans during their execution,

    M. Beetz and D. McDermott, “Improving robot plans during their execution,” inProc. 2nd Int. Conf. Artif. Intell. Planning Syst. (AIPS), 1994, pp. 3–12

    1994

  27. [27]

    Plan execution monitoring and control architecture for mobile robots,

    F. Noreils and R. Chatila, “Plan execution monitoring and control architecture for mobile robots,”IEEE Trans. Robot. Autom., vol. 11, no. 2, pp. 255–266, 1995

    1995

  28. [28]

    Intelligent execution monitoring in dynamic environments,

    M. Fichtner, A. Großmann, and M. Thielscher, “Intelligent execution monitoring in dynamic environments,”Fundam. Inform., vol. 57, no. 2–4, pp. 371–392, 2003

    2003

  29. [29]

    RePLan: Robotic replanning with perception and language models,

    M. Skretaet al., “RePLan: Robotic replanning with perception and language models,”arXiv:2401.04157, 2024

    Pith/arXiv arXiv 2024

This paper was first reviewed by deepseek-v4-flash on August 2, 2026.