Pith. sign in

REVIEW 2 major objections 2 minor 53 references

Autonomous Obstacle Removal for Excavators through Policy Learning with Particle Simulation

T0 review · 2 major / 2 minor · reviewed 2026-06-27 · grok-4.3

Pith's one-line read A burial-depth curriculum in particle simulation trains an excavator policy for obstacle removal that transfers successfully to a real 12-ton machine.

desk verdict The burial-depth curriculum lets them train an excavator policy in particle sim that transfers to a real 12-ton machine, but the abstract gives almost no numbers and no check on whether the sim matches real soil behavior. read the letter →

arxiv 2606.09183 v1 pith:XKCRVYZD submitted 2026-06-08 cs.RO

classification cs.RO
keywords obstacleremovalexcavatorpolicylearningparticlesimulationcurriculumsim-to-realroboticsearthwork
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

The paper aims to automate repeated excavation cycles for removing obstacles whose burial conditions change over time. It argues that particle-based simulation can capture the needed soil-obstacle contacts, terrain changes, and visibility, but only if computational cost is managed. The proposed approach uses a curriculum that begins with shallow burials and steadily increases depth while lowering particle count, thereby raising difficulty and lowering simulation expense in tandem. The resulting policy takes RGB-D measurements as input and produces parameterized trajectories as output, using the identical interface in simulation and on the physical excavator. A sympathetic reader would care because the method shows how to make long-horizon robotic earthmoving policies trainable at all.

What carries the argument

The burial-conditioned curriculum that varies obstacle burial depth to control both task difficulty and the number of particles required for accurate simulation.

What would settle it

Direct real-world trials in which the transferred policy fails to remove obstacles that the simulated policy removed under matching burial depths and obstacle placements would falsify the transfer claim.

Watch

Extended reading notes

Core claim

The paper claims that a time-efficient sim-to-real policy learning framework using particle simulation under controllable burial conditions, together with a curriculum that progressively increases burial depth while adjusting particle count, produces an effective obstacle-removal policy within three days. The policy observes terrain and obstacle information from RGB-D measurements and outputs a parameterized excavation trajectory. Experiments show this policy transfers to a real 12-ton excavator operating on open ground with various steel obstacles, while baseline methods fail even after a full week of training.

Load-bearing premise

The particle simulation must accurately reproduce real-world soil-obstacle interactions including contact states, terrain deformation, and obstacle visibility under the burial conditions used in training.

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, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 2 minor

Summary. The paper proposes a burial-conditioned curriculum learning framework for training excavation policies in particle-based simulation. The policy takes RGB-D observations of terrain and obstacles and outputs parameterized trajectories. The curriculum starts with shallow burials (low particle count, low difficulty) and progressively increases burial depth while scaling particle count to control both task difficulty and simulation cost. Experiments claim the approach learns an effective policy in three days of training, while baselines fail after a full week, and that the resulting policy transfers successfully to a physical 12-ton excavator performing obstacle removal on open ground with steel obstacles.

Significance. If the sim-to-real transfer claim holds, the work demonstrates a practical route to learning state-dependent excavation policies that adapt to changing soil-obstacle conditions, a long-standing barrier in autonomous earthmoving. The curriculum that jointly modulates burial depth and particle count is a concrete engineering contribution that addresses the computational cost of repeated-cycle particle simulation. The manuscript supplies no machine-checked proofs or parameter-free derivations, but the reproducible observation-action interface between simulator and real excavator is a strength that supports the transfer narrative.

major comments (2)
  1. [Abstract / Experiments] Abstract and Experiments section: The central claim that the curriculum 'successfully learns an effective obstacle-removal policy' and 'achieves successful transfer' is presented without any quantitative metrics, success rates, error bars, statistical tests, or definition of how success was measured (e.g., number of cycles to removal, failure modes). This absence prevents evaluation of the performance gap versus baselines.
  2. [Simulation / Transfer] Simulation and Transfer sections: The sim-to-real claim rests on the unvalidated assumption that the particle simulator reproduces real soil-obstacle contact states, terrain deformation, and obstacle visibility across the burial depths used in the curriculum. No side-by-side quantitative validation (deformation maps, contact force metrics, or burial-resistance curves) is reported; any systematic mismatch would mean the policy exploits simulator artifacts rather than learning transferable behavior.
minor comments (2)
  1. [Abstract] The abstract states that 'baseline methods fail even after a full week of training' without specifying the baseline algorithms, their hyper-parameters, or training details; this information belongs in the Experiments section for reproducibility.
  2. [Methods] Notation for the parameterized trajectory (e.g., what parameters are output by the policy) is introduced without an explicit equation or diagram in the methods description.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and will revise the manuscript to incorporate quantitative metrics and clarify sim-to-real aspects where possible.

read point-by-point responses
  1. Referee: [Abstract / Experiments] Abstract and Experiments section: The central claim that the curriculum 'successfully learns an effective obstacle-removal policy' and 'achieves successful transfer' is presented without any quantitative metrics, success rates, error bars, statistical tests, or definition of how success was measured (e.g., number of cycles to removal, failure modes). This absence prevents evaluation of the performance gap versus baselines.

    Authors: We agree that the abstract and experiments lack explicit quantitative metrics. In the revised version we will add success rates (percentage of successful removals over repeated trials), definition of success (e.g., obstacle fully removed within a maximum number of cycles), failure modes, error bars from multiple random seeds, and statistical comparisons against baselines. These additions will appear in both the abstract and the experiments section. revision: yes

  2. Referee: [Simulation / Transfer] Simulation and Transfer sections: The sim-to-real claim rests on the unvalidated assumption that the particle simulator reproduces real soil-obstacle contact states, terrain deformation, and obstacle visibility across the burial depths used in the curriculum. No side-by-side quantitative validation (deformation maps, contact force metrics, or burial-resistance curves) is reported; any systematic mismatch would mean the policy exploits simulator artifacts rather than learning transferable behavior.

    Authors: The manuscript reports no quantitative side-by-side validation of simulator outputs against real soil measurements. We will revise the relevant sections to explicitly state this limitation, describe the reproducible observation-action interface used for transfer, and present the real-world excavator results as the primary empirical evidence of transfer. We will also add any available qualitative comparisons if they exist in our data. revision: partial

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: empirical sim-to-real policy learning with no self-referential derivations

full rationale

The paper presents an empirical framework for training an excavation policy in particle simulation under a burial-depth curriculum, then transferring to a physical excavator. Claims rest on experimental outcomes (successful learning in 3 days vs. baselines failing after a week, plus real-world transfer) rather than any derivation chain. No equations, fitted parameters renamed as predictions, or load-bearing self-citations appear in the provided text; the particle simulator is treated as an external training environment whose fidelity is an empirical precondition, not a constructed result. This matches the default expectation of a non-circular empirical robotics paper.

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

Abstract provides insufficient technical detail to enumerate specific free parameters, axioms, or invented entities; the approach implicitly relies on the fidelity of particle simulation for soil dynamics.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Autonomous Obstacle Removal for Excavators through Policy Learning with Particle Simulation." pith.science (2026). https://pith.science/paper/XKCRVYZD

@misc{pith2026260609183,
  author       = {Pith},
  title        = {Pith review of: Autonomous Obstacle Removal for Excavators through Policy Learning with Particle Simulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XKCRVYZD}},
  note         = {Machine review of arXiv:2606.09183}
}
read the original abstract

Autonomous obstacle removal from the ground is an important earthwork task, but this is difficult to automate because an excavator must adapt its excavation trajectories over repeated cycles as soil-obstacle conditions change. Learning such state-dependent behavior requires a training environment that reproduces accumulated soil-obstacle interactions, including contact states, terrain deformation, and obstacle visibility. Accordingly, particle-based simulation is suitable for the relevant policy learning. However, particle simulation is computationally expensive, and repeated excavation cycles further increase the learning cost. We observe that the burial condition of an obstacle governs both task difficulty and simulation cost: deeper burial makes obstacle removal harder while also requiring more particles for accurate simulation. This observation motivates a burial-conditioned curriculum learning strategy. We propose a time-efficient sim-to-real policy learning framework in which the policy observes terrain and obstacle information from RGB-D measurements and then outputs a parameterized excavation trajectory; in this process, the simulator reproduces in a real-world excavator the same observation-action interface it uses under controllable burial conditions. The curriculum begins with shallow burial conditions and progressively increases burial depth while adjusting particle count, thus simultaneously controlling task difficulty and simulation cost. Experiments show that the proposed framework successfully learns an effective obstacle-removal policy, whereas baseline methods fail even after a full week of training. The proposed curriculum achieves effective performance within three days and achieves successful transfer to a real 12-ton excavator operating on open ground with various steel obstacles, thus demonstrating robust obstacle removal.

Figures

Figures reproduced from arXiv: 2606.09183 by the authors.

Figure 1
Figure 1. Overview of proposed sim-to-real policy learning framework for [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of real-world observation–action interface. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Overview of policy learning framework in particle simulation. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (9 more)
Figure 5
Figure 5. Figure 5: Comparison of action and observation designs. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Comparison of task success rates in sim-to-sim and sim-to-real experiments. In real-world environment, I-shaped and L-shaped obstacles are used, [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Analysis of policy behavior. Snapshots illustrate the environment during task execution. The overlaid 3D heatmap visualizes the observed depth [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Comparison of task completion time between learned policy and [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 9
Figure 9. Figure 9: Application demonstrations of proposed obstacle-removal policy. [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 10
Figure 10. Figure 10: Detailed definition of trajectory parameterization. Action parameters [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]
Figure 11
Figure 11. Figure 11: Qualitative comparison of predefined excavation behaviors between real-world environment and simulation environment. [PITH_FULL_IMAGE:figures/full_fig_p013_11.png]
Figure 12
Figure 12. Figure 12: Comparison of sample efficiency between designed actions and fine [PITH_FULL_IMAGE:figures/full_fig_p014_12.png]
Figure 13
Figure 13. Figure 13: Comparison of the effects of domain randomization parameters on success rates. Excavation success rates for seven obstacle shapes in Curriculum 3 [PITH_FULL_IMAGE:figures/full_fig_p015_13.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 2 canonical work pages

  1. [1]

    Robotic Autonomous Systems for Earthmoving Equipment Operating in V olatile Conditions and Teaming Capacity: a Survey,

    H. A. Nguyen and Q. P. Ha, “Robotic Autonomous Systems for Earthmoving Equipment Operating in V olatile Conditions and Teaming Capacity: a Survey,”Robotica, vol. 41, no. 2, pp. 486–510, 2023

  2. [2]

    Sensing, Perception, Decision, Planning and Action of Autonomous Excavators,

    O. M. U. Eraliev, K.-H. Lee, D.-Y . Shin, and C.-H. Lee, “Sensing, Perception, Decision, Planning and Action of Autonomous Excavators,” Automation in Construction, vol. 141, p. 104428, 2022

  3. [3]

    An Autonomous Excavator System for Material Loading Tasks,

    L. Zhang, J. Zhao, P. Long, L. Wang, L. Qian, F. Lu, X. Song, and D. Manocha, “An Autonomous Excavator System for Material Loading Tasks,”Science Robotics, vol. 6, no. 55, 2021

  4. [4]

    Autonomous Free- Form Trenching Using a Walking Excavator,

    D. Jud, P. Leemann, S. Kerscher, and M. Hutter, “Autonomous Free- Form Trenching Using a Walking Excavator,”IEEE Robotics and Automation Letters, vol. 4, no. 4, pp. 3208–3215, 2019

  5. [5]

    Excavation learning for rigid objects in clutter,

    Q. Lu and L. Zhang, “Excavation learning for rigid objects in clutter,” IEEE Robotics and Automation Letters, vol. 6, no. 4, pp. 7373–7380, 2021

  6. [6]

    Tasknet: A neural task planner for autonomous excavator,

    J. Zhao and L. Zhang, “Tasknet: A neural task planner for autonomous excavator,” inIEEE International Conference on Robotics and Automa- tion, 2021, pp. 2220–2226

  7. [7]

    Learning excavation of rigid objects with offline reinforcement learning,

    S. Jin, Z. Ye, and L. Zhang, “Learning excavation of rigid objects with offline reinforcement learning,”arXiv preprint arXiv:2303.16427, 2023

  8. [8]

    Excavation Reinforcement Learning Us- ing Geometric Representation,

    Q. Lu, Y . Zhu, and L. Zhang, “Excavation Reinforcement Learning Us- ing Geometric Representation,”IEEE Robotics and Automation Letters, vol. 7, no. 2, pp. 4472–4479, 2022

Show all 53 references
  1. [9]

    Simulation-Based Reinforcement Learning Ap- proach Towards Construction Machine Automation,

    K. Matsumoto, A. Yamaguchi, T. Oka, M. Yasumoto, S. Hara, M. Iida, and M. Teichmann, “Simulation-Based Reinforcement Learning Ap- proach Towards Construction Machine Automation,” inProceedings of the International Symposium on Automation and Robotics in Construc- tion, vol. 37...

  2. [10]

    Time-Jerk Optimal Trajectory Planning of Hydraulic Robotic Excavator,

    Y . Zhang, Z. Sun, Q. Sun, Y . Wang, X. Li, and J. Yang, “Time-Jerk Optimal Trajectory Planning of Hydraulic Robotic Excavator,”Advances in Mechanical Engineering, vol. 13, no. 7, pp. 1–13, 2021. 16

  3. [11]

    Optimization-based Framework for Excavation Trajectory Generation,

    Y . Yang, P. Long, X. Song, J. Pan, and L. Zhang, “Optimization-based Framework for Excavation Trajectory Generation,”IEEE Robotics and Automation Letters, vol. 6, no. 2, pp. 1479–1486, 2021

  4. [12]

    Modeling and Offset-free Model Predictive Control of a Hydraulic Mini Excavator,

    F. A. Bender, S. G ¨oltz, T. Br ¨aunl, and O. Sawodny, “Modeling and Offset-free Model Predictive Control of a Hydraulic Mini Excavator,” IEEE Transactions on Automation Science and Engineering, vol. 14, no. 4, pp. 1682–1694, 2017

  5. [13]

    Three-Dimensionsal Granular Flow Continuum Modeling via Material Point Method with Hyperelastic Nonlocal Granular Fluidity,

    A. Haeri and K. Skonieczny, “Three-Dimensionsal Granular Flow Continuum Modeling via Material Point Method with Hyperelastic Nonlocal Granular Fluidity,”Computer Methods in Applied Mechanics and Engineering, vol. 394, p. 114904, 2022

  6. [14]

    Progressive-Resolution Policy Distillation: Leveraging Coarse-Resolution Simulations for Time- Efficient Fine-Resolution Policy Learning,

    Y . Kadokawa, H. Tahara, and T. Matsubara, “Progressive-Resolution Policy Distillation: Leveraging Coarse-Resolution Simulations for Time- Efficient Fine-Resolution Policy Learning,”IEEE Transactions on Au- tomation Science and Engineering, vol. 22, pp. 18 682–18 693, 2025

  7. [15]

    Granular Gym: High Performance Simulation for Robotic Tasks with Granular Materials,

    D. Millard, D. Pastor, J. Bowkett, P. Backes, and G. S. Sukhatme, “Granular Gym: High Performance Simulation for Robotic Tasks with Granular Materials,”arXiv preprint arXiv:2306.01369, 2023

  8. [16]

    Design of Highly Realistic Virtual Environment for Excavator Simulator,

    T. Ni, H. Zhang, C. Yu, D. Zhao, and S. Liu, “Design of Highly Realistic Virtual Environment for Excavator Simulator,”Computers & Electrical Engineering, vol. 39, no. 7, pp. 2112–2123, 2013

  9. [17]

    A Survey on Curriculum Learn- ing,

    X. Wang, Y . Chen, and W. Zhu, “A Survey on Curriculum Learn- ing,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 9, pp. 4555–4576, 2022

  10. [18]

    Curriculum Learning for Reinforcement Learning Domains: A Framework and Survey,

    S. Narvekar, B. Peng, M. Leonetti, J. Sinapov, M. E. Taylor, and P. Stone, “Curriculum Learning for Reinforcement Learning Domains: A Framework and Survey,”Journal of Machine Learning Research, vol. 21, no. 181, pp. 1–50, 2020

  11. [19]

    Autonomous Loading System for Load-Haul-Dump (LHD) Machines Used in Underground Mining,

    C. Tampier, M. Mascaro, and J. Ruiz-del Solar, “Autonomous Loading System for Load-Haul-Dump (LHD) Machines Used in Underground Mining,”Applied Sciences, vol. 11, no. 18, p. 8718, 2021

  12. [20]

    Admittance Control for Robotic Loading: Design and Experiments with a 1-Tonne Loader and a 14-Tonne Load-Haul-Dump Machine,

    A. A. Dobson, J. A. Marshall, and J. Larsson, “Admittance Control for Robotic Loading: Design and Experiments with a 1-Tonne Loader and a 14-Tonne Load-Haul-Dump Machine,”Journal of Field Robotics, vol. 34, no. 1, pp. 123–150, 2017

  13. [21]

    Autonomous Excavation of Rocks Using a Gaussian Process Model and Unscented Kalman Filter,

    F. E. Sotiropoulos and H. H. Asada, “Autonomous Excavation of Rocks Using a Gaussian Process Model and Unscented Kalman Filter,”IEEE Robotics and Automation Letters, vol. 5, no. 2, pp. 2491–2497, 2020

  14. [22]

    Real-Time Motion Planning of a Hydraulic Excavator Using Trajectory Optimization and Model Predictive Control,

    D. Lee, I. Jang, J. Byun, H. Seo, and H. J. Kim, “Real-Time Motion Planning of a Hydraulic Excavator Using Trajectory Optimization and Model Predictive Control,” inIEEE/RSJ International Conference on Intelligent Robots and Systems, 2021, pp. 2135–2142

  15. [23]

    Expert-Emulating Excavation Trajectory Planning for Autonomous Robotic Industrial Excavator,

    B. Son, C. Kim, C. Kim, and D. Lee, “Expert-Emulating Excavation Trajectory Planning for Autonomous Robotic Industrial Excavator,” in IEEE/RSJ International Conference on Intelligent Robots and Systems, 2020, pp. 2656–2662

  16. [24]

    Deep Learning-based Autonomous Excavation: a Bucket-trajectory Planning Algorithm,

    J. Huh, J. Bae, D. Lee, J. Kwak, C. Moon, C. Im, Y . Ko, T. K. Kang, and D. Hong, “Deep Learning-based Autonomous Excavation: a Bucket-trajectory Planning Algorithm,”IEEE Access, vol. 11, pp. 38 047–38 060, 2023

  17. [25]

    Particle Finite Element Method Implementation for Large Deformation Analysis Using Abaqus,

    W.-H. Yuan, H.-C. Wang, W. Zhang, B.-B. Dai, K. Liu, and Y . Wang, “Particle Finite Element Method Implementation for Large Deformation Analysis Using Abaqus,”Acta Geotechnica, vol. 16, no. 8, pp. 2449– 2462, 2021

  18. [26]

    Excavation of fragmented rocks with multi-modal model-based reinforcement learning,

    Y . Zhu, L. Wang, and L. Zhang, “Excavation of fragmented rocks with multi-modal model-based reinforcement learning,” inIEEE/RSJ International Conference on Intelligent Robots and Systems, 2022, pp. 6523–6530

  19. [27]

    The material point method for simulating continuum materials,

    C. Jiang, C. Schroeder, A. Selle, J. Teran, and A. Stomakhin, “The material point method for simulating continuum materials,” inInterna- tional Conference and Exhibition on Computer Graphics and Interactive Techniques, 2016, pp. 1–52

  20. [28]

    A Novel Robust Imitation Learning Framework for Complex Skills with Limited Demonstrations,

    W. Wang, C. Zeng, H. Zhan, and C. Yang, “A Novel Robust Imitation Learning Framework for Complex Skills with Limited Demonstrations,” IEEE Transactions on Automation Science and Engineering, vol. 22, pp. 3947–3959, 2025

  21. [29]

    Imitation Learning: A Survey of Learning Methods,

    A. Hussein, M. M. Gaber, E. Elyan, and C. Jayne, “Imitation Learning: A Survey of Learning Methods,”ACM Computing Surveys, vol. 50, no. 2, pp. 1–35, 2017

  22. [30]

    Disturbance-Injected Robust Imitation Learning with Task Achieve- ment,

    H. Tahara, H. Sasaki, H. Oh, B. Michael, and T. Matsubara, “Disturbance-Injected Robust Imitation Learning with Task Achieve- ment,” inInternational Conference on Robotics and Automation, 2022, pp. 2466–2472

  23. [31]

    Cyclic Policy Distillation: Sample-Efficient Sim-to-Real Reinforcement Learning with Domain Randomization,

    Y . Kadokawa, L. Zhu, Y . Tsurumine, and T. Matsubara, “Cyclic Policy Distillation: Sample-Efficient Sim-to-Real Reinforcement Learning with Domain Randomization,”Robotics and Autonomous Systems, vol. 165, 2023

  24. [32]

    Learning Robotic Powder Weighing from Simulation for Laboratory Automation,

    Y . Kadokawa, M. Hamaya, and K. Tanaka, “Learning Robotic Powder Weighing from Simulation for Laboratory Automation,” inIEEE/RSJ International Conference on Intelligent Robots and Systems, 2023, pp. 2932–2939

  25. [33]

    Teacher-Student Curriculum Learning,

    T. Matiisen, A. Oliver, T. Cohen, and J. Schulman, “Teacher-Student Curriculum Learning,”IEEE Transactions on Neural Networks and Learning Systems, vol. 31, no. 9, pp. 3732–3740, 2020

  26. [34]

    Sim-to- Real Transfer of Robotic Control with Dynamics Randomization,

    X. B. Peng, M. Andrychowicz, W. Zaremba, and P. Abbeel, “Sim-to- Real Transfer of Robotic Control with Dynamics Randomization,” in IEEE International Conference on Robotics and Automation, 2018, pp. 3803–3810

  27. [35]

    Randomized ensembled double q-learning: Learning fast without a model,

    X. Chen, C. Wang, Z. Zhou, and K. Ross, “Randomized ensembled double q-learning: Learning fast without a model,” inInternational Conference on Learning Representations, 2021

  28. [36]

    Research on an Open Source Physical Simulator for Autonomous Construction Machinery Development,

    D. Endo, Y . Matsusaka, G. Yamauchi, and T. Hashimoto, “Research on an Open Source Physical Simulator for Autonomous Construction Machinery Development,” inInternational Symposium on Automation and Robotics in Construction, 2024, pp. 1303–1306

  29. [37]

    Robotic manipulation and sensing of deformable objects in domestic and in- dustrial applications: A survey,

    J. Sanchez, J.-A. Corrales, B.-C. Bouzgarrou, and Y . Mezouar, “Robotic manipulation and sensing of deformable objects in domestic and in- dustrial applications: A survey,”The International Journal of Robotics Research, vol. 37, no. 7, pp. 688–716, 2018

  30. [38]

    Diffsrl: Learning dynamical state representation for deformable object manipulation with differentiable simulation,

    S. Chen, Y . Liu, S. W. Yao, J. Li, T. Fan, and J. Pan, “Diffsrl: Learning dynamical state representation for deformable object manipulation with differentiable simulation,”IEEE Robotics and Automation Letters, vol. 7, no. 4, pp. 9533–9540, 2022

  31. [39]

    Prolonging tool life: Learning skillful use of general-purpose tools through lifespan- guided reinforcement learning,

    P.-Y . Wu, C.-Y . Kuo, Y . Kadokawa, and T. Matsubara, “Prolonging tool life: Learning skillful use of general-purpose tools through lifespan- guided reinforcement learning,”IEEE Access, vol. 14, pp. 24 715– 24 731, 2026

  32. [40]

    Visuo- motor Reinforcement Learning for Multirobot Cooperative Navigation,

    Z. Liu, Q. Liu, L. Tang, K. Jin, H. Wang, M. Liu, and H. Wang, “Visuo- motor Reinforcement Learning for Multirobot Cooperative Navigation,” IEEE Transactions on Automation Science and Engineering, vol. 19, no. 4, pp. 3234–3245, 2022

  33. [41]

    HEAP—The Autonomous Walking Exca- vator,

    D. Jud, S. Kerscher, M. Wermelinger, E. Jelavic, P. Egli, P. Leemann, G. Hottiger, and M. Hutter, “HEAP—The Autonomous Walking Exca- vator,”Automation in Construction, vol. 129, p. 103783, 2021

  34. [42]

    Effects of machine instability feedback on safety during digging operation in teleoperated excavators,

    M. Ito, C. Raima, S. Saiki, Y . Yamazaki, and Y . Kurita, “Effects of machine instability feedback on safety during digging operation in teleoperated excavators,”IEEE Access, vol. 9, pp. 28 987–28 998, 2021

  35. [43]

    Integrated task and motion planning,

    C. R. Garrett, R. Chitnis, R. Holladay, B. Kim, T. Silver, L. P. Kaelbling, and T. Lozano-P’erez, “Integrated task and motion planning,”Annual Review of Control, Robotics, and Autonomous Systems, vol. 4, pp. 265– 293, 2021

  36. [44]

    Robotic embankment: Free-form autonomous formation in terrain with HEAP,

    D. Jud, I. Hurkxkens, C. Girot, and M. Hutter, “Robotic embankment: Free-form autonomous formation in terrain with HEAP,”Construction Robotics, vol. 5, no. 2, pp. 101–113, 2021

  37. [45]

    Faster R-CNN: Towards real- time object detection with region proposal networks,

    S. Ren, K. He, R. Girshick, and J. Sun, “Faster R-CNN: Towards real- time object detection with region proposal networks,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, no. 6, pp. 1137– 1149, 2017

  38. [46]

    Towards open world object detection,

    K. J. Joseph, S. Khan, F. S. Khan, and V . N. Balasubramanian, “Towards open world object detection,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 5830–5840

  39. [47]

    Robust object detection under occlusion with context-aware CompositionalNets,

    A. Wang, Y . Sun, A. Kortylewski, and A. L. Yuille, “Robust object detection under occlusion with context-aware CompositionalNets,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 12 645–12 654

  40. [48]

    Deep visual foresight for planning robot motion,

    C. Finn and S. Levine, “Deep visual foresight for planning robot motion,” inIEEE International Conference on Robotics and Automation, 2017, pp. 2786–2793

  41. [49]

    de Boor,A Practical Guide to Splines

    C. de Boor,A Practical Guide to Splines. Springer, 1978, vol. 27

  42. [50]

    A note on piecewise linear and multilinear table interpolation in many dimensions,

    A. Weiser and S. E. Zarantonello, “A note on piecewise linear and multilinear table interpolation in many dimensions,”Mathematics of Computation, vol. 50, no. 181, pp. 189–196, 1988

  43. [51]

    A multiscale retinex for bridging the gap between color images and the human observation of scenes,

    D. J. Jobson, Z.-u. Rahman, and G. A. Woodell, “A multiscale retinex for bridging the gap between color images and the human observation of scenes,”IEEE Transactions on Image Processing, vol. 6, no. 7, pp. 965–976, 1997

  44. [52]

    Bilateral filtering for gray and color images,

    C. Tomasi and R. Manduchi, “Bilateral filtering for gray and color images,” inProceedings of the IEEE International Conference on Computer Vision, 1998, pp. 839–846

  45. [53]

    ”GrabCut

    C. Rother, V . Kolmogorov, and A. Blake, “”GrabCut”: Interactive foreground extraction using iterated graph cuts,”ACM Transactions on Graphics, vol. 23, no. 3, pp. 309–314, 2004

Pith tools

Reviewed June 27, 2026 · model on record in the stance chip above.