Pith. sign in

REVIEW 2 major objections 5 minor 1 cited by

MOSAIC: Skill-Centric Manipulation Planning with Physics Simulation

T0 review · 2 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read MOSAIC plans long-horizon manipulation by focusing search on regions where skills are demonstrably effective, composing imperfect skills without symbolic task specifications.

desk verdict MOSAIC's oracle and generator/connector split are a real step for composing imperfect skills, but the probabilistic completeness proof is a sketch that does not survive contact with continuous parameter spaces and the algorithm's own nearest-neighbor node selection. read the letter →

arxiv 2504.16738 v3 pith:W5LPZ2NX submitted 2025-04-23 cs.RO cs.AI

Itamar Mishani , Yorai Shaoul , Maxim Likhachev This is my paper · ORCID
classification cs.ROcs.AI
keywords skill-centricplanninglong-horizonmanipulationphysicssimulationgeneratorandconnectorskillsmosaicgraphprobabilisticcompletenesstaskmotiondiffusionpolicy
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

MOSAIC is a planning framework for long-horizon robot manipulation that composes a library of imperfect, generic skills—such as pushing, grasping, and transporting—without requiring a symbolic task specification. Its central claim is that planning should be anchored in 'islands of competence': regions of the state space where a skill is demonstrably likely to succeed, discovered by rolling skills out in a physics simulator during search. A multi-directional search builds a graph of skill trajectories, with generator skills producing local trajectories and connector skills linking them by solving boundary value problems. The paper argues this feasibility-driven approach solves tasks whose solutions require non-obvious intermediate steps, and that it outperforms sequential skill chaining and roadmap baselines in success rate, generalization, and scalability. If true, it would let robots handle novel tabletop tasks by composing existing skills rather than learning new policies or manually engineering symbolic planners.

What carries the argument

The central mechanism is the mosaic graph together with the two skill families. Generators are parameterized skills that produce trajectories without specified start or goal states, creating new nodes; connectors are conditional skills that produce trajectories between specified boundary conditions, creating edges. A third component, the oracle, orchestrates search: it decides whether to invoke a generator or connector based on the node-to-edge ratio, selects skills by a score balancing past success rate with an exploration bonus, and chooses node pairs to connect via random, goal-bias, start-bias, or unification modes. Physics simulation is used to evaluate every skill rollout, so only valid, collision-free trajectories enter the graph, and the fraction of invalid rollouts in a batch is reflected in the trajectory's cost. This machinery shifts the search from goal-directed to multi-directional and feasibility-anchored.

What would settle it

Run MOSAIC's planned skill sequences for the three tabletop scenarios on the physical robot and compare per-skill success rates with the same sequences executed in the simulator. If the real-world success rate for pushes or grasps in the discovered islands falls far below the simulated rate, or if the planner must re-plan after contact failures on hardware, the competence-island transfer premise is falsified.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that long-horizon manipulation planning can be recast as graph construction over skill trajectories, where the graph is grown not from the start or goal but from wherever the skills themselves are reliable. The core discovery is that imperfect skills can be made into reliable building blocks by using physics simulation as an in-the-loop oracle: each candidate trajectory is rolled out to verify contacts and prune infeasible plans, and the success statistics guide further exploration. The paper calls the resulting object a mosaic graph—nodes are generator-produced trajectories, edges are connector trajectories that satisfy boundary conditions—and shows that a domain-independent statistical oracle that balances exploitation of successful skills with exploration of underused ones can build this graph efficiently. It also proves probabilistic completeness for the skill-centric setting under deterministic skill-trajectory generation: given infinite iterations, every skill and parameter combination is eventually invoked, so any feasible solution decomposable into library skills is found.

Load-bearing premise

The whole approach rests on the physics simulator predicting real contact outcomes well enough that the competence islands found in simulation remain competent on the real robot.

Editorial extensions

If this is right

  • Long-horizon manipulation problems can be solved without hand-written symbolic preconditions and effects, since the simulator supplies feasibility checks and boundary conditions are defined by world states.
  • Skill libraries with imperfect components, such as a push policy with roughly 70% success, can be composed into reliable plans by concentrating effort on regions where the skills succeed.
  • The framework's multi-directional search finds solutions involving non-obvious intermediate steps, such as clearing a movable obstacle before grasping, which directed forward or backward search tends to miss.
  • Probabilistic completeness guarantees that if a feasible decomposition into the given skill library exists, MOSAIC will find it asymptotically.
  • Because the oracle is domain-independent, the same algorithm and skill library can be reused across tasks without task-specific tuning.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A testable extension is to swap the statistical oracle for a large language model or learned affordance model; the paper notes the oracle can be any module, so an LLM that proposes connector boundary conditions could reduce search on tasks with strong semantic structure.
  • If competence islands are task-independent, the same generator-generated nodes could be cached and reused across planning queries, turning MOSAIC into an anytime roadmap that amortizes simulation cost over repeated tasks.
  • The sim-to-real dependence suggests MOSAIC's practical ceiling tracks simulator fidelity; using a learned world model or real-to-sim calibration in place of the physics engine would change the competence estimates and is a direct way to test the framework's robustness.
  • The probabilistic-completeness argument assumes deterministic skill-trajectory generation; extending it to stochastic policies would require a different convergence notion, since batch rollouts only provide empirical success probabilities.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 5 minor

Summary. This paper proposes MOSAIC, a multi-directional, skill-centric planning algorithm for long-horizon manipulation. It constructs a graph whose nodes are generated skill trajectories and whose edges are connector-skill trajectories, using an oracle to choose skills, sample parameters, and select node pairs to connect, with physics simulation used to evaluate skill outcomes. The paper claims probabilistic completeness for the framework and reports experiments on three tabletop manipulation scenarios in simulation and on a real UR10e, claiming superior performance over skill-chaining, CEM, and roadmap-based baselines.

Significance. The core idea of focusing planning on 'islands of competence' and using physics simulation to prune infeasible skill compositions is a plausible and practically useful contribution to skill-based manipulation planning. If the claims are substantiated, the framework would offer a way to plan with imperfect learned skills without hand-coded symbolic preconditions and effects. The paper also provides detailed descriptions of the oracle, skill library, and experimental setups. However, the formal probabilistic-completeness proof is not valid as written, and the real-world validation is only qualitative, so the strongest theoretical and empirical claims outrun the evidence.

major comments (2)
  1. [IV-C (Definition 4 and proof sketch)] The proof that MOSAIC is probabilistically complete is not correct. Condition (1) requires invoking every generator with 'every parameter configuration' and condition (2) requires attempting to connect 'every disconnected node' with every connector and parameter. The oracle samples parameters from continuous distributions (e.g., object poses and diffusion seeds), so any particular exact parameter configuration has probability zero of being sampled; 'all parameter configurations' cannot be covered by random sampling without additional dense-sampling and clearance assumptions, which are not stated. More importantly, the oracle's node-pair selection in Appendix S-A2 only connects a node to its nearest neighbor (or to start/goal-biased nearest neighbors) and never systematically attempts arbitrary disconnected node pairs. A feasible solution whose intermediate trajectory segments correspond to two nodes that are not nearest neighbors can therefore be missed forever, directly contradicting condition (2). The assertion that nonzero selection probabilities guarantee eventual exploration of all skill/parameter and node-pair configurations conflates 'nonzero probability of selecting a skill' with 'coverage of uncountable configuration spaces and arbitrary node pairs.' I recommend either providing a rigorous proof under explicit dense-sampling and node-pair coverage assumptions or replacing the probabilistic-completeness claim with a weaker, precisely stated completeness or consistency guarantee that actually follows from the algorithm.
  2. [V-D (real-world validation)] The abstract and contributions state that MOSAIC demonstrates efficacy 'in both simulation and the real world,' but the real-world evidence in Section V-D consists solely of the sentence 'Our hardware validation tests (Fig. 2) trended similarly; View videos in our supplementary materials.' No quantitative hardware results are reported: no number of trials, success/failure counts, execution times, or failure analysis. Because real-world performance is explicitly claimed as part of the contribution, either report the hardware results in a quantitative form or revise the claim to a qualitative validation.
minor comments (5)
  1. [V-D / Fig. 3] The contribution statement claims that MOSAIC 'outperforms existing methods in performance, generalization, and scalability,' but the experiments do not directly measure generalization (e.g., novel object poses or unseen clutter configurations) or scalability (e.g., increasing number of objects or skills). Consider adding explicit tests or qualifying the claim.
  2. [Appendix S-A2 (Oracle node-pair selection)] The informal description says the oracle 'restricts the available skills to be only the connector skills if there are fewer nodes than vertices in the mosaic graph.' Since the text defines V(E) as the number of edges, this likely should read 'fewer edges than nodes'; as written the sentence is confusing.
  3. [V-C (CEM baseline)] The Cross-Entropy Method baseline is described as evaluating skill-parameter sequences 'based on task progress' and selecting top candidates, but the precise progress/score function used for ranking is not specified. This makes the comparison difficult to reproduce and interpret.
  4. [IV-C (Definition 4)] The definition of Π* as 'the set of all feasible solutions' is informal; in a rigorous probabilistic-completeness statement the solution set should be defined precisely, particularly whether it is finite or infinite and how 'discovered' trajectories at iteration k are measured.
  5. [V / Appendix S-B] The values and initialization of the oracle parameters α, p_lb, p_up, p_s, p_g, and p_{s-g} are not reported. Please state whether these were fixed across all scenarios and how they were chosen, especially given the claim that the oracle 'requires no task-specific tuning.'

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: MOSAIC's central derivation and benchmark claims do not reduce to their own inputs, and the self-citations are not load-bearing.

full rationale

The paper's derivation chain is self-contained against its stated model. The oracle selects skills and node pairs from graph statistics; skill parameters are sampled randomly; and every generated trajectory is evaluated by forward simulation in Sapien, not by the goal predicate or by the final benchmark outcome (Section IV-B, Section V-A). The success rates used by the oracle are empirical rollout statistics obtained during search, not constants fitted to the experimental results. The probabilistic-completeness argument in Section IV-C is an asymptotic coverage claim based on non-zero selection probabilities and random sampling; whether it is fully valid for continuous parameter spaces and nearest-neighbor-only node pairing is a correctness concern, not a circularity, because the conclusion is not assumed by the premises. The related-work citations to prior work by the same authors ([7], [9], [51]) motivate physics-based simulation and diffusion-policy skills but do not provide the central theorem or the experimental outcomes, and no uniqueness theorem or ansatz is imported from those citations. The claim that MOSAIC 'outperforms existing methods' is supported by the independent baseline comparisons in Section V-D, not by an equation that equates the output with an input. Consequently, no circular step can be exhibited from the text.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The paper introduces algorithmic concepts (generators, connectors, oracle) but no new physical entities, forces, or conserved quantities. The central claim rests on the hand-chosen oracle hyperparameters, simulator fidelity, and the asserted exploration properties, which are the main ledgers beyond the skill library itself.

free parameters (3)
  • alpha (oracle exploitation-exploration weight) = not specified
    Appears in the oracle skill-selection score U(sigma_i) = alpha*s_σ + (1-alpha)*sqrt(ln(sum(t+1))/(t+1)) + n (Section IV-B). Hand-chosen; no ablation or sensitivity analysis is reported.
  • Node/edge ratio thresholds p_lb, p_up = not specified
    Control the probability of selecting connector skills based on the edge-to-node ratio (Section VII-A). Hand-chosen.
  • Node-pair selection cutoffs p_s, p_g, p_{s-g} = not specified
    Determine the randomization mode (START, GOAL, START-GOAL, RANDOM) in the oracle (Section VII-A). Hand-chosen.
assumptions (4)
  • domain assumption The physics simulator (Sapien) accurately predicts real-world contact outcomes for the skills used.
    Section V-A states skills are rolled out in Sapien to verify contact outcomes. Real-world validation is qualitative only ('trended similarly', Section V-D), so this fidelity is assumed, not measured.
  • domain assumption Skill trajectories are valid: collision-free and dynamically feasible.
    Section III states 'Skills produce only valid trajectories tau_i: not causing collisions between the robot and static obstacles and respecting robot dynamics.'
  • ad hoc to paper The oracle's exploration assigns non-zero probability to every skill and parameter configuration, including all connector attempts.
    Section IV-C uses this to argue probabilistic completeness. With continuous parameter spaces and nearest-neighbor-only node-pair selection (Section VII-A), this is asserted but not proved.
  • domain assumption A feasible solution exists in the given skill library for the task.
    Probabilistic completeness (Definition 4) is conditional on the existence of a feasible solution; MOSAIC only guarantees finding a solution that exists in the library.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MOSAIC: Skill-Centric Manipulation Planning with Physics Simulation." pith.science (2026). https://pith.science/paper/W5LPZ2NX

@misc{pith2026250416738,
  author       = {Pith},
  title        = {Pith review of: MOSAIC: Skill-Centric Manipulation Planning with Physics Simulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W5LPZ2NX}},
  note         = {Machine review of arXiv:2504.16738}
}
read the original abstract

Planning long-horizon manipulation motions using a set of predefined skills is a central challenge in robotics; solving it efficiently could enable general-purpose robots to tackle novel tasks by flexibly composing generic skills. Solutions to this problem lie in an infinitely vast space of parameterized skill sequences -- a space where common incremental methods struggle to find sequences that have non-obvious intermediate steps. Some approaches reason over lower-dimensional, symbolic spaces, which are more tractable to explore but may be brittle and are laborious to construct. In this work, we introduce MOSAIC, a skill-centric, multi-directional planning approach that targets these challenges by reasoning about which skills to employ and where they are most likely to succeed, by utilizing physics simulation to estimate skill execution outcomes. Specifically, MOSAIC employs two complementary skill families: Generators, which identify ``islands of competence'' where skills are demonstrably effective, and Connectors, which link these skill-trajectories by solving boundary value problems. By focusing planning efforts on regions of high competence, MOSAIC efficiently discovers physically-grounded solutions. We demonstrate its efficacy on complex long-horizon problems in both simulation and the real world, using a diverse set of skills including generative diffusion models, motion planning algorithms, and manipulation-specific models. Visit skill-mosaic.github.io for demonstrations and examples.

Figures

Figures reproduced from arXiv: 2504.16738 by the authors.

Figure 1
Figure 1. MOSAIC solves long-horizon manipulation tasks by generating local skill trajectories (circles) and connecting those with connector skills (squares). MOSAIC capitalizes on the skills themselves to guide the exploration process toward regions where they are likely to succeed – enabling effective composition of generic local skills to solve complex tasks. Abstract— Planning long-horizon manipulation motions us￾ing a se… view at source ↗
Figure 2
Figure 2. Simulation setups and their real-world counterparts. Across all scenarios, the robot must place the plate into the bin. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Algorithms comparison across experimental scenarios. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Deliberate Practice: Learning Robot Skills under a Budget

    cs.RO 2026-08 conditional novelty 7.0 of 10

    Deliberate Practice allocates a limited robot practice budget across skills by solving an exact bilinear program, and beats greedy active-learning baselines in long-horizon manipulation tasks.

Reference graph

Works this paper leans on

61 extracted references · 52 canonical work pages · cited by 1 Pith paper

  1. [1]

    Isaac Sim

    NVIDIA, “Isaac Sim.”

  2. [2]

    Genesis: A universal and generative physics engine for robotics and beyond,

    G. Authors, “Genesis: A universal and generative physics engine for robotics and beyond,” 2024

    work page 2024

  3. [3]

    SAPIEN: A simulated part-based interactive environment,

    F. Xiang, Y . Qin, K. Mo, Y . Xia, H. Zhu, F. Liu, M. Liu, H. Jiang, Y . Yuan, H. Wang, L. Yi, A. X. Chang, L. J. Guibas, and H. Su, “SAPIEN: A simulated part-based interactive environment,” inThe IEEE Con- ference on Computer Vision and Pattern Recognition (CVPR), 2020

    work page 2020

  4. [4]

    Mujoco: A physics engine for model-based control,

    E. Todorov, T. Erez, and Y . Tassa, “Mujoco: A physics engine for model-based control,” in2012 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2012, pp. 5026–5033

    2012

  5. [5]

    Mujoco playground,

    K. Zakka, B. Tabanpour, Q. Liao, M. Haiderbhai, S. Holt, J. Y . Luo, A. Allshire, E. Frey, K. Sreenath, L. A. Kahrset al., “Mujoco playground,”arXiv preprint arXiv:2502.08844, 2025

    arXiv 2025

  6. [6]

    Reconciling reality through simulation: A real-to-sim-to-real approach for robust manipulation,

    M. Torne, A. Simeonov, Z. Li, A. Chan, T. Chen, A. Gupta, and P. Agrawal, “Reconciling reality through simulation: A real-to-sim-to-real approach for robust manipulation,”arXiv preprint arXiv:2403.03949, 2024

    arXiv 2024

  7. [7]

    Planning with selec- tive physics-based simulation for manipulation among movable objects,

    M. S. Saleem and M. Likhachev, “Planning with selec- tive physics-based simulation for manipulation among movable objects,” in2020 IEEE International Confer- ence on Robotics and Automation (ICRA), 2020, pp. 6752–6758

    work page 2020

  8. [8]

    Manipulation planning among movable obstacles us- ing physics-based adaptive motion primitives,

    D. M. Saxena, M. S. Saleem, and M. Likhachev, “Manipulation planning among movable obstacles us- ing physics-based adaptive motion primitives,” in2021 IEEE International Conference on Robotics and Au- tomation (ICRA), 2021, pp. 6570–6576

    work page 2021

Show all 61 references
  1. [9]

    Planning for manipu- lation among movable objects: Deciding which objects go where, in what order, and how,

    D. Saxena and M. Likhachev, “Planning for manipu- lation among movable objects: Deciding which objects go where, in what order, and how,” inProceedings of the International Conference on Automated Planning and Scheduling, vol. 33, 2023, pp. 668–676

    2023

  2. [10]

    Long-horizon torque-limited planning through contact using discrete search and continuous optimization,

    R. Natarajan, G. L. Johnston, N. Simaan, M. Likhachev, and H. Choset, “Long-horizon torque-limited planning through contact using discrete search and continuous optimization,” inIROS 2023 Workshop on Leveraging Models for Contact-Rich Manipulation, 2023

    2023

  3. [11]

    Fast contact-implicit model predictive control,

    S. Le Cleac’h, T. A. Howell, S. Yang, C.-Y . Lee, J. Zhang, A. Bishop, M. Schwager, and Z. Manchester, “Fast contact-implicit model predictive control,”IEEE Transactions on Robotics, 2024

    2024

  4. [12]

    Model tensor planning,

    A. T. Le, K. Nguyen, M. N. Vu, J. Carvalho, and J. Peters, “Model tensor planning,”arXiv preprint arXiv:2505.01059, 2025

    2025 arXiv

  5. [13]

    Dif- ferentiable gpu-parallelized task and motion planning,

    W. Shen, C. Garrett, N. Kumar, A. Goyal, T. Hermans, L. P. Kaelbling, T. Lozano-P ´erez, and F. Ramos, “Dif- ferentiable gpu-parallelized task and motion planning,” 2025

    2025

  6. [14]

    Robotic world model: A neural network simulator for robust policy opti- mization in robotics,

    C. Li, A. Krause, and M. Hutter, “Robotic world model: A neural network simulator for robust policy opti- mization in robotics,”arXiv preprint arXiv:2501.10100, 2025

    2025

  7. [15]

    V-jepa 2: Self-supervised video models enable understanding, prediction and planning,

    M. Assran, A. Bardes, D. Fan, Q. Garrido, R. Howes, M. Muckley, A. Rizvi, C. Roberts, K. Sinha, A. Zholus et al., “V-jepa 2: Self-supervised video models enable understanding, prediction and planning,”arXiv preprint arXiv:2506.09985, 2025

    2025 arXiv

  8. [16]

    Integrated task and motion planning,

    C. R. Garrett, R. Chitnis, R. Holladay, B. Kim, T. Sil- ver, L. P. Kaelbling, and T. Lozano-P ´erez, “Integrated task and motion planning,”Annual review of control, robotics, and autonomous systems, 2021

    2021

  9. [17]

    Planning for robots with skills,

    M. Crosby, F. Rovida, M. R. Pedersen, R. P. Petrick, and V . Kr¨uger, “Planning for robots with skills,” in4th ICAPS Workshop on Planning and Robotics, 2016

    2016

  10. [18]

    From skills to symbols: Learning symbolic represen- tations for abstract high-level planning,

    G. Konidaris, L. P. Kaelbling, and T. Lozano-Perez, “From skills to symbols: Learning symbolic represen- tations for abstract high-level planning,”Journal of Artificial Intelligence Research, 2018

    2018

  11. [19]

    Pdsketch: Integrated domain programming, learning, and planning,

    J. Mao, T. Lozano-P ´erez, J. Tenenbaum, and L. Kael- bling, “Pdsketch: Integrated domain programming, learning, and planning,”Advances in Neural Informa- tion Processing Systems, 2022

    2022

  12. [20]

    Guiding search in continuous state-action spaces by learning an action sampler from off-target search experience,

    B. Kim, L. Kaelbling, and T. Lozano-P ´erez, “Guiding search in continuous state-action spaces by learning an action sampler from off-target search experience,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 32, no. 1, 2018

    2018

  13. [21]

    Learning compositional models of robot skills for task and motion planning,

    Z. Wang, C. R. Garrett, L. P. Kaelbling, and T. Lozano- P´erez, “Learning compositional models of robot skills for task and motion planning,”The International Jour- nal of Robotics Research, 2021

    2021

  14. [22]

    Sym- bolic state estimation with predicates for contact-rich manipulation tasks,

    T. Migimatsu, W. Lian, J. Bohg, and S. Schaal, “Sym- bolic state estimation with predicates for contact-rich manipulation tasks,” in2022 International Conference on Robotics and Automation (ICRA), 2022

    2022

  15. [23]

    Learning neuro- symbolic skills for bilevel planning,

    T. Silver, A. Athalye, J. B. Tenenbaum, T. Lozano- P´erez, and L. P. Kaelbling, “Learning neuro- symbolic skills for bilevel planning,”arXiv preprint arXiv:2206.10680, 2022

    2022 arXiv

  16. [24]

    Search-based task planning with learned skill effect models for lifelong robotic manip- ulation,

    J. Liang, M. Sharma, A. LaGrassa, S. Vats, S. Saxena, and O. Kroemer, “Search-based task planning with learned skill effect models for lifelong robotic manip- ulation,” in2022 International Conference on Robotics and Automation (ICRA), 2022, pp. 6351–6357

    2022

  17. [25]

    Dimsam: Diffusion mod- els as samplers for task and motion planning under partial observability,

    X. Fang, C. R. Garrett, C. Eppner, T. Lozano-P ´erez, L. P. Kaelbling, and D. Fox, “Dimsam: Diffusion mod- els as samplers for task and motion planning under partial observability,” in2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2024, pp. 1412–1419

    2024

  18. [26]

    Spire: Synergistic planning, imitation, and re- inforcement learning for long-horizon manipulation,

    Z. Zhou, A. Garg, D. Fox, C. Garrett, and A. Man- dlekar, “Spire: Synergistic planning, imitation, and re- inforcement learning for long-horizon manipulation,” arXiv preprint arXiv:2410.18065, 2024

    2024 arXiv

  19. [27]

    Plan- ning with diffusion for flexible behavior synthesis,

    M. Janner, Y . Du, J. Tenenbaum, and S. Levine, “Plan- ning with diffusion for flexible behavior synthesis,” in International Conference on Machine Learning, 2022

    2022

  20. [28]

    Diffusion policy: Vi- suomotor policy learning via action diffusion,

    C. Chi, Z. Xu, S. Feng, E. Cousineau, Y . Du, B. Burch- fiel, R. Tedrake, and S. Song, “Diffusion policy: Vi- suomotor policy learning via action diffusion,”The International Journal of Robotics Research, 2024

    2024

  21. [29]

    Au- tomatic goal generation for reinforcement learning agents,

    C. Florensa, D. Held, X. Geng, and P. Abbeel, “Au- tomatic goal generation for reinforcement learning agents,” inInternational conference on machine learn- ing, 2018, pp. 1515–1528

    2018

  22. [30]

    Intrinsically motivated goal exploration processes with automatic curriculum learning,

    S. Forestier, R. Portelas, Y . Mollard, and P.-Y . Oudeyer, “Intrinsically motivated goal exploration processes with automatic curriculum learning,”Journal of Machine Learning Research, vol. 23, no. 152, pp. 1–41, 2022

    2022

  23. [31]

    Dis- covery of options via meta-learned subgoals,

    V . Veeriah, T. Zahavy, M. Hessel, Z. Xu, J. Oh, I. Ke- maev, H. P. van Hasselt, D. Silver, and S. Singh, “Dis- covery of options via meta-learned subgoals,”Advances in Neural Information Processing Systems, vol. 34, pp. 29 861–29 873, 2021

    2021

  24. [32]

    Skill discovery in con- tinuous reinforcement learning domains using skill chaining,

    G. Konidaris and A. Barto, “Skill discovery in con- tinuous reinforcement learning domains using skill chaining,”Advances in neural information processing systems, vol. 22, 2009

    2009

  25. [33]

    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, 1999

    1999

  26. [34]

    Option discovery using deep skill chaining,

    A. Bagaria and G. Konidaris, “Option discovery using deep skill chaining,” inInternational Conference on Learning Representations, 2019

    2019

  27. [35]

    Robustly learning composable options in deep rein- forcement learning,

    A. Bagaria, J. Senthil, M. Slivinski, and G. Konidaris, “Robustly learning composable options in deep rein- forcement learning,” inProceedings of the 30th Interna- tional Joint Conference on Artificial Intelligence, 2021

    2021

  28. [36]

    Skill discovery for exploration and planning using deep skill graphs,

    A. Bagaria, J. K. Senthil, and G. Konidaris, “Skill discovery for exploration and planning using deep skill graphs,” inInternational Conference on Machine Learning, 2021, pp. 521–531

    2021

  29. [37]

    Deep affordance foresight: Planning through what can be done in the future,

    D. Xu, A. Mandlekar, R. Mart ´ın-Mart´ın, Y . Zhu, S. Savarese, and L. Fei-Fei, “Deep affordance foresight: Planning through what can be done in the future,” in 2021 IEEE international conference on robotics and automation (ICRA), 2021, pp. 6206–6213

    2021

  30. [38]

    Augmenting rein- forcement learning with behavior primitives for diverse manipulation tasks,

    S. Nasiriany, H. Liu, and Y . Zhu, “Augmenting rein- forcement learning with behavior primitives for diverse manipulation tasks,” inInternational Conference on Robotics and Automation (ICRA), 2022

    2022

  31. [39]

    Gener- ative skill chaining: Long-horizon skill planning with diffusion models,

    U. A. Mishra, S. Xue, Y . Chen, and D. Xu, “Gener- ative skill chaining: Long-horizon skill planning with diffusion models,” inConference on Robot Learning, 2023

    2023

  32. [40]

    Generative fac- tor chaining: Coordinated manipulation with diffusion- based factor graph,

    U. A. Mishra, Y . Chen, and D. Xu, “Generative fac- tor chaining: Coordinated manipulation with diffusion- based factor graph,” in8th Annual Conference on Robot Learning, 2024

    2024

  33. [41]

    Roadmaps with gaps over controllers: Achieving efficiency in planning under dynamics,

    A. Sivaramakrishnan, S. Tangirala, E. Granados, N. R. Carver, and K. E. Bekris, “Roadmaps with gaps over controllers: Achieving efficiency in planning under dynamics,” inIEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2024

    2024

  34. [42]

    Reset- free lifelong learning with skill-space planning,

    K. Lu, A. Grover, P. Abbeel, and I. Mordatch, “Reset- free lifelong learning with skill-space planning,”arXiv preprint arXiv:2012.03548, 2020

    2012 arXiv

  35. [43]

    Learning symbolic representations for planning with parameter- ized skills,

    B. Ames, A. Thackston, and G. Konidaris, “Learning symbolic representations for planning with parameter- ized skills,” inIEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2018

    2018

  36. [44]

    Rrt-connect: An efficient approach to single-query path planning,

    J. Kuffner and S. LaValle, “Rrt-connect: An efficient approach to single-query path planning,”IEEE Interna- tional Conference on Robotics and Automation, 2000

    2000

  37. [45]

    The open motion planning library,

    I. A. Sucan, M. Moll, and L. E. Kavraki, “The open motion planning library,”IEEE Robotics and Automa- tion Magazine, pp. 72–82, 2012

    2012

  38. [46]

    Maniskill2: A unified benchmark for generalizable manipulation skills,

    J. Gu, F. Xiang, X. Li, Z. Ling, X. Liu, T. Mu, Y . Tang, S. Tao, X. Wei, Y . Yao, X. Yuan, P. Xie, Z. Huang, R. Chen, and H. Su, “Maniskill2: A unified benchmark for generalizable manipulation skills,” inInternational Conference on Learning Representations, 2023

    2023

  39. [47]

    R. Y . Rubinstein and D. P. Kroese,The cross-entropy method: a unified approach to combinatorial optimiza- tion, Monte-Carlo simulation and machine learning. Springer Science & Business Media, 2004

    2004

  40. [48]

    Probabilistic roadmaps for path planning in high-dimensional configuration spaces,

    L. Kavraki, P. Svestka, J.-C. Latombe, and M. Over- mars, “Probabilistic roadmaps for path planning in high-dimensional configuration spaces,”IEEE Transac- tions on Robotics and Automation, pp. 566–580, 1996

    1996

  41. [49]

    A note on two problems in connexion with graphs,

    E. DIJKSTRA, “A note on two problems in connexion with graphs,”Numerische Mathematik, 1959

    1959

  42. [50]

    Motion planning diffusion: Learning and planning of robot motions with diffusion models,

    J. Carvalho, A. T. Le, M. Baierl, D. Koert, and J. Peters, “Motion planning diffusion: Learning and planning of robot motions with diffusion models,” inInternational Conference on Intelligent Robots and Systems, 2023

    2023

  43. [51]

    Multi-robot motion planning with diffusion models,

    Y . Shaoul, I. Mishani, S. Vats, J. Li, and M. Likhachev, “Multi-robot motion planning with diffusion models,” 2024. VII. APPENDIX In this appendix we will provide additional details about various practical components that we presented in this work. We include specifics about ...

    2024

  44. [52]

    First (potentially) restricting the library of skills to a subset and then choosing a particular skill within it

    Skill Selection:The oracle chooses skills to invoke in two stages. First (potentially) restricting the library of skills to a subset and then choosing a particular skill within it. Skill Type Selection: To decide whether to invoke a generator skill (to add nodes to the mosaic ...

  45. [53]

    Node Pair Selection for Connection:Another decision made by the oracle module is deciding which pairs of nodes in the mosaic graph are worthwhile to attempt connecting. Since the mosaic graph is initially comprised of disconnected components representing different regions in t...

  46. [54]

    Random selection: selecting a random node and con- necting it to its nearest neighbor

  47. [55]

    Goal bias: if any nodes in the mosaic graph are known to be connected to a terminal node (i.e., a node whose trajectory terminates at a state satisfying the goal con- dition), choose one at random and attempt to connect it to its nearest neighbor

  48. [56]

    Start bias: among all nodes in the mosaic graph that are connected to the start node (i.e., the node holding the start state singleton trajectory), choose one at random and attempt connecting it to its nearest neighbor

  49. [57]

    Unification bias: if nodes connected to both the start and a terminal node exist, select a pair of these for a connection attempt. A note on node pair selection.To avoid repeated un- successful connection attempts, the oracle assigns penalties to node pairs with repeated faile...

  50. [58]

    The generator push skill uses adiffusionpush policy to generate random object pushes on the tabletop

    Push Skill:In this work, we define two types ofPush skills: generator and connector. The generator push skill uses adiffusionpush policy to generate random object pushes on the tabletop. Its parameters include the object’s start and goal poses, as well as a seed (within a pred...

  51. [59]

    Pick Skill, Score-based Geometric Antipodal Grasp: The pick skill’s policy begins with selecting an antipodal grasp that maximizes a score function while ensuring a valid inverse kinematics (IK) solution, followed by a screw-based motion planner that moves the robot through a ...

  52. [60]

    Sampling Antipodal Grasp Candidates.From the pro- cessed point cloud, a set of antipodal grasp candidates is generated based on the object’s axis-aligned bounding box (AABB)

    and fitting a local plane to these neighbors via Principal Component Analysis (PCA), where the normal is defined as the eigenvector corresponding to the smallest eigenvalue. Sampling Antipodal Grasp Candidates.From the pro- cessed point cloud, a set of antipodal grasp candidat...

  53. [61]

    If no such sequence is found, the algorithm reports failure

    is used to compute a sequence of skill trajectories that connect them. If no such sequence is found, the algorithm reports failure. Incremental MOSAIC-Roadmapis our closest baseline to MOSAIC. Its operation is generally similar to MOSAIC- Roadmap, with the main difference betw...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.