Pith. sign in

REVIEW 4 major objections 7 minor 39 references

CDM: Contact Diffusion Model for Multi-Contact Point Localization

T0 review · 4 major / 7 minor · reviewed 2026-08-08 · deepseek-v4-flash

Pith's one-line read The paper's central claim is that a diffusion model conditioned on joint torques, base wrench, past outputs, and a signed distance field can localize one or two contact points, including cases where several contact/force pairs produce…

desk verdict A genuinely new diffusion-based approach to multi-contact localization with real-robot evidence; the headline error metric flatters the method, but the core contribution is solid. read the letter →

arxiv 2502.06109 v1 pith:5NANKDD6 submitted 2025-02-10 cs.RO

classification cs.RO
keywords contactlocalizationdiffusionmodelmulti-contactproprioceptivesensingsigneddistancefieldrobotmanipulationmulti-modalposteriordenoisingprobabilistic
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

CDM treats contact localization as posterior sampling rather than point estimation: given a window of joint-torque and base-wrench observations, the model draws 64 candidate contact points from the distribution conditioned on those observations and on the signed distance field of the robot's surfaces. The paper's central claim is that this generative formulation resolves the singularity in which multiple pairs of contact points and forces produce exactly the same sensor readings, because the samples can cover all modes instead of collapsing to one. A second claim is that conditioning on the model's own earlier output—freezing the first contact once a classifier says only one contact is present—reduces multi-modality when a second contact appears sequentially. If these claims hold, multi-contact localization no longer needs the threshold tuning that particle-filter methods require, and one model handles single-, transition-, and steady-dual-contact cases. The paper reports real-world matched RMSE of 0.44 cm for single contact and 1.24 cm for dual contact, at 15.97 ms per inference.

What carries the argument

The load-bearing object is the conditional denoising diffusion model with a FiLM (feature-wise linear modulation) conditioned MLP denoiser. At each reverse step $k$, the denoiser predicts the noise given a window of proprioceptive observations, the saved first-contact estimate $\hat{X}_{T_s}$ (or null when no first contact is known), the diffusion step $k$, and, for each candidate point, the signed distance and gradient to every link surface from a pretrained SDF network. These conditions enter through FiLM layers: one per-point affine transform from the SDF feature and one shared affine transform from a global observation feature, so the surface geometry directly reshapes the denoising trajectory. A separate contact-state classifier, trained with cross-entropy loss, decides whether the current window is single-contact; if it is, the historical first contact is frozen to the current output. The training loss is the standard DDPM noise-prediction loss, with an extra random diffusion step applied to the first contact point so the network learns with and without historical information.

What would settle it

Run the trained model on a trajectory where a single contact slides along the arm after the classifier has frozen the first-contact estimate; if the frozen $\hat{X}_{T_s}$ does not track the sliding point, the fixed-first-contact assumption fails and the real-world M-RMSE should exceed the reported 1.24 cm dual-contact error.

Watch

Extended reading notes

Core claim

The discovery is that the ill-posed mapping from contact points and forces to sensor measurements, which leaves many equally consistent explanations, is better handled by sampling the posterior than by optimizing a single estimate. CDM uses a denoising diffusion network whose reverse process is conditioned on a window of proprioceptive observations, the current joint configuration, and the signed distance field (SDF) of every link, and it outputs a set of points that jointly cover the true contact locations. The contact-state classifier identifies whether the current window is single-contact; if so, the first-contact estimate is saved and used as an additional condition for subsequent windows, so a sequentially appearing second contact is localized relative to an already-fixed first contact. In simulation the method reaches M-RMSE (distance from each true contact to the nearest generated sample) of 0.32 cm for single contact and 1.14 cm for steady dual contact; in real-world tests it reaches 0.44 cm and 1.24 cm, respectively, using ten denoising steps at 15.97 ms.

Load-bearing premise

The load-bearing premise is that there are at most two contacts, each is a pure point contact transmitting no moment, the first contact stays fixed once detected, and the momentum-observer wrench estimate is unbiased; if any of these fails, the conditioned posterior and the frozen historical contact are wrong.

Editorial extensions

If this is right

  • One model covers single-contact, transition dual-contact, and steady dual-contact states, eliminating the user-tuned threshold used by particle-filter methods to decide when a new contact appears.
  • Because the output is a set of samples, singular cases with several valid contact positions remain represented instead of collapsing to one point estimate, so downstream planning can act on the full uncertainty.
  • The 15.97 ms inference time for ten denoising steps is constant regardless of contact count and is fast enough for closed-loop contact-aware control.
  • SDF conditioning keeps generated points on the robot surface: the average distance between generated samples and the surface is 0.29 cm with SDF versus 0.89 cm without.
  • Historical conditioning more than halves M-RMSE and cluster-center error in sequential dual-contact localization compared with null conditioning (M-RMSE 0.62 cm versus 1.33 cm in simulation).

Reading between the lines

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

  • An implication the authors leave implicit: the same posterior-sampling formulation could be extended to contact-force identification, because the measurement model is linear in the forces once the contact geometry is sampled; the paper lists this as future work.
  • A testable extension beyond the paper: because the model transfers from simulation to hardware without fine-tuning, its checkpoints could be evaluated on a different arm of similar kinematics after recomputing the SDF, which would separate the method's generality from its training distribution.
  • An editorial observation: the reported single-contact accuracy relies on an extra base F/T sensor that earlier learning-based comparisons may lack, so the 0.44 cm figure likely mixes the value of the generative model with the value of that additional sensing modality.
  • A possible consequence for control: the 64-sample posterior could be summarized by its modes and fed to contact-aware model-predictive control, turning the multi-modal distribution into robust constraint sets rather than a single assumed contact.
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

4 major / 7 minor

Summary. The paper proposes CDM, a conditional diffusion model for localizing up to two sequential contact points on a robot arm using proprioceptive sensing (joint torque sensors and a base F/T sensor). The model is conditioned on a window of past observations, its own previous single-contact output, and a signed distance field of the robot surface, and it generates a 64-point cloud representing the posterior over contact locations. Training uses simulation data with a 90/10 split; experiments include simulation metrics, ablations on SDF and historical conditioning, and direct sim-to-real transfer on a Kinova Gen3 arm with 100 single-contact and 100 dual-contact trials. The abstract claims real-world localization errors of 0.44 cm (single) and 1.24 cm (dual) at 15.97 ms inference time.

Significance. If the quantitative claims are properly interpreted, CDM is a novel and promising application of generative modeling to contact localization: it is, to my knowledge, the first diffusion-based contact localizer, it handles multi-modality explicitly, and the real-robot dual-contact evaluation is a useful addition to a literature that has mostly evaluated in simulation. The 90/10 held-out simulation split, the direct real-robot transfer test, and the ablation studies are genuine strengths. The main weakness is that the headline accuracy metric is not the error of a localized point, so the claimed 'high accuracy' overstates what the system delivers for downstream point-estimate tasks.

major comments (4)
  1. [Section IV-A, Table III, Abstract] The headline real-world errors (0.44 cm single, 1.24 cm dual) are M-RMSE values, defined as (1/nc) * sum_j min_{xi in X̂_t} ||r_t,j - xi||. This metric only asks whether each true contact has at least one nearby sample among the 64 generated points; it does not report the error of any localized point estimate. The same trials in Table III give C-RMSE-1 = 0.68 cm for single contact and C-RMSE-1/2 = 2.61/2.85 cm for dual contact, i.e., the cluster-center errors are roughly 2–2.3 times larger than the abstract's figures. Since 'localization' conventionally means producing a specific point, and downstream controllers need a point estimate, the abstract's use of M-RMSE as 'the error' materially overstates the method's accuracy. Please report C-RMSE (or another pointwise error) as the headline accuracy, and state explicitly what M-RMSE does and does not certify.
  2. [Table III and Table II] The real-world results in Table III report only mean values over 100 trials, with no standard deviation, confidence interval, or distribution information, despite the robot configuration being randomized every 20 experiments. Table II likewise gives no variance for the 614,400 simulation runs. Without dispersion measures, the reader cannot assess the reliability of the claimed 'high accuracy' or the significance of the 40% improvement over the PF baseline [12]. Please include variance statistics or, at minimum, report per-trial error distributions (e.g., median and interquartile range) for the real-world experiments.
  3. [Algorithm 2 and Section III.A] Algorithm 2 freezes X̂_Ts once the contact state classifier labels the current observation as single-contact, and it never re-estimates X̂_Ts during subsequent dual-contact phases. This design assumes the first contact remains stationary after it appears. If the first contact slides or rolls during the dual-contact phase, or if the classifier mislabels a steady dual-contact as single, the historical conditioning is based on a stale first-contact location. The assumptions in Section III.A (point contact without moment, up to two sequential contacts) are stated, but the consequence of first-contact motion is not discussed or evaluated. Please add an explicit limitation statement and, ideally, an experiment with a moving first contact to delimit the range of validity of the central claim.
  4. [Section IV-A, Fig. 4, Table III] The paper's central contribution is the prediction of a multi-modal posterior, yet no quantitative metric evaluates whether the generated distribution is calibrated or whether the number and location of modes are correct. M-RMSE is a coverage metric that ignores spurious modes, and C-RMSE forces k-means clustering with the number of clusters equal to the true contact count, which cannot detect extra or missing modes. Since the singularity-handling claim rests on the model producing the correct multi-modal structure, please add distribution-level evaluation, such as mode count accuracy, coverage of high-probability regions, or a proper scoring rule for the empirical posterior.
minor comments (7)
  1. [Section II.C] The section title contains a typo: 'Denosing diffusion probabilistic models' should read 'Denoising diffusion probabilistic models.'
  2. [Algorithm 1] The loss in Algorithm 1 is written as Lθ = ||ϵ - ϵθ(...)||, omitting the square that appears in Eq. (6); please make the two consistent.
  3. [Section III.B] The dataset notation D = {ξ_i}_{N_d}^{i=1} is unconventional; it should be written as D = {ξ_i}_{i=1}^{N_d}.
  4. [Section IV.D, Table III] Footnote 1 correctly notes that direct comparison with [20] and [21] is complicated by the additional base F/T sensor, but the main text still states 'the proposed method achieves higher accuracy than existing learning-based methods.' Please qualify this comparison in the main text as well, not only in the footnote.
  5. [Section III.D / Fig. 3] The SDF network fφ is described as pre-trained and taken from [30], but no details are given about its training data, resolution, or whether it is fine-tuned during CDM training. Please clarify, since the SDF conditioning is a central architectural component.
  6. [Table I] The contact-state classifier failure rates are reported per state, but the label definition for 'trans-dual' (nc = 1 → 2) versus 'steady dual' (nc = 2) is not explicit in Section III.B. Please define precisely what labels gψ is trained to predict for each of the three states.
  7. [Section IV.A] The equation for M-RMSE is given inline without a number, which makes it awkward to reference in the text and in tables; please number it.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the learned posterior is trained on held-out simulation and validated on real hardware, with no fitted parameter renamed as prediction.

full rationale

The paper's central claim is that a conditional diffusion model approximates the posterior p(X_t | O_{t-T:t}, X_Ts, S) for contact localization. This is a learned mapping, not a derivation from Eq. (1). Eq. (1) is the physical forward model used only to generate the simulation dataset and to define the QP used as an evaluation metric; Section II.B explicitly states that the QP 'is not used in the proposed method, it is employed in Section IV-A to evaluate the points generated by CDM.' The conditioning on past diffusion outputs (Algorithm 2) is a recursive inference scheme whose inputs are previous model estimates, not the ground-truth targets of the current timestep, and the ablation in Table II compares CDM-his with CDM-null on the same data; no parameter is fit to those targets and then reported as a prediction. The self-citation to the authors' prior particle-filter work [12] is used as a comparison baseline and as motivation, not as an unverified premise that forces the result. The SDF conditioning uses an externally published method [30]. The M-RMSE metric in Tables I and III is a best-of-64 containment measure, and the abstract's 'error of 1.24 cm' is therefore an optimistic summary of dual-contact accuracy; but that is an evaluation-transparency concern, not a circularity of derivation, since M-RMSE does not enter training or the model's construction. The derivation chain is self-contained: simulation-trained, held-out evaluation, and real-world deployment without retuning.

Assumptions & free parameters 6 free parameters · 7 assumptions · 0 invented entities

The central claim rests on a learned mapping from sensor windows to contact-point sets, so most explanatory power sits in the fitted network parameters. The domain assumptions (point contact, nc <= 2, fixed first contact, accurate DOB and SDF) are stated but not independently validated; together they bound when the method can be trusted.

free parameters (6)
  • Denoiser network parameters (143M) = learned, not enumerated
    Fitted by minimizing the DDPM noise-prediction loss on 90% of the simulation dataset; these weights directly determine where generated contact points are placed.
  • Contact state classifier parameters (15M) = learned, not enumerated
    Fitted with cross-entropy to label single/transition/steady-dual contact states; errors affect historical conditioning.
  • SDF network parameters (from Deep Medial Fields) = pretrained, not enumerated
    Provides signed distance and surface direction for every link; the paper relies on its accuracy to keep samples on the surface.
  • Observation window length T = 60 ms
    Hand-chosen; limits the posterior to a finite memory. If contacts change faster than T, the conditioning loses information.
  • Number of generated points np = 64
    Chosen as a speed/coverage trade-off; M-RMSE is computed as min over these points, so np affects the reported error.
  • Diffusion steps K and noise schedule = K=1000, beta_1=1e-6, beta_K=1e-3, 10 DDIM steps at inference
    Standard DDPM hyperparameters chosen for this problem; inference-time step count affects both speed and sample quality.
assumptions (7)
  • domain assumption Pure point contact with no transmitted moment (Eq. 1).
    Section II.A: the measurement model sums Ai(q,ri) Fi, so contacts cannot transmit torques; real contacts with soft materials or large patches violate this.
  • domain assumption At most two contacts, occurring sequentially, with up to one per link.
    Section III.A: the posterior (8) is restricted to nc <= 2 and one contact already present at Ts; three simultaneous contacts are out of scope.
  • domain assumption The momentum-based disturbance observer provides an unbiased estimate of external wrench.
    Section II.A: (1) is evaluated with W_hat_ext from the DOB; biased estimates directly feed the conditioning and the QP error metric.
  • domain assumption The first contact point remains fixed once it appears.
    Section III.C.1 and Algorithm 2: historical conditioning freezes X_Ts during dual-contact phases; sliding or moving first contacts are not modeled.
  • domain assumption The SDF network accurately represents the robot surface at arbitrary query points.
    Section III.C.2: f_phi is used as a conditioning input to pull points toward the surface; the ablation shows 0.29 cm average distance with SDF, so accuracy is assumed.
  • domain assumption The training distribution covers the real-world deployment distribution.
    Section IV: random MuJoCo configurations with forces 10-25 N and sensor noise are used to train; real-world contacts must lie in this distribution for sim-to-real transfer to hold.
  • standard math DDPM's reverse process approximates the posterior p(Xt | conditions).
    Section II.C: relies on the standard DDPM formulation (Eqs. 3-6) to treat 64 generated samples as samples from the conditional posterior.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CDM: Contact Diffusion Model for Multi-Contact Point Localization." pith.science (2026). https://pith.science/paper/5NANKDD6

@misc{pith2026250206109,
  author       = {Pith},
  title        = {Pith review of: CDM: Contact Diffusion Model for Multi-Contact Point Localization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5NANKDD6}},
  note         = {Machine review of arXiv:2502.06109}
}
read the original abstract

In this paper, we propose a Contact Diffusion Model (CDM), a novel learning-based approach for multi-contact point localization. We consider a robot equipped with joint torque sensors and a force/torque sensor at the base. By leveraging a diffusion model, CDM addresses the singularity where multiple pairs of contact points and forces produce identical sensor measurements. We formulate CDM to be conditioned on past model outputs to account for the time-dependent characteristics of the multi-contact scenarios. Moreover, to effectively address the complex shape of the robot surfaces, we incorporate the signed distance field in the denoising process. Consequently, CDM can localize contacts at arbitrary locations with high accuracy. Simulation and real-world experiments demonstrate the effectiveness of the proposed method. In particular, CDM operates at 15.97ms and, in the real world, achieves an error of 0.44cm in single-contact scenarios and 1.24cm in dual-contact scenarios.

Figures

Figures reproduced from arXiv: 2502.06109 by the authors.

Figure 1
Figure 1. The objective of this paper is to localize the contact points [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The dataset is collected from scenarios where dual-contact occurs [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. In the inner loop, the denoiser sequentially eliminates noise from the white Gaussian noise (from [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Experiment #1: The generated samples from CDM, corresponding to different contact states: (a) Single-contact. (b) Single-contact, but the posterior (9) has two modes. (c) Transition dual-contact. (d) Steady dual￾contact. (i) Failure rate: We assess the contact state cl…
Figure 5
Figure 5. Figure 5: Experiment #3: The samples generated by CDM: (a) with historical result conditioning, and (b) without it. D. Direct sim-to-real transfer In experiment #4, we validate CDM using a real-world robot (see [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 30 canonical work pages

  1. [12]

    Proprioceptive sensor-based simultaneous multi-contact point localization and force identification for robotic arms,

    S. W. Han and M. J. Kim, “Proprioceptive sensor-based simultaneous multi-contact point localization and force identification for robotic arms,” in 2023 IEEE International Conference on Robotics and Automation (ICRA), pp. 12 099–12 105

  2. [1]

    A reachability tree-based algorithm for robot task and motion planning,

    K. Kim, D. Park, and M. J. Kim, “A reachability tree-based algorithm for robot task and motion planning,” in 2023 IEEE International Conference on Robotics and Automation (ICRA) , pp. 3750–3756

  3. [2]

    Diffusion policy: Visuomotor policy learning via action diffusion,

    C. Chi, S. Feng, Y . Du, Z. Xu, E. Cousineau, B. Burchfiel, and S. Song, “Diffusion policy: Visuomotor policy learning via action diffusion,” arXiv preprint arXiv:2303.04137 , 2023

  4. [3]

    Disturbance observer with constraints,

    T. H. Yun and M. J. Kim, “Disturbance observer with constraints,” IEEE Control Systems Letters , vol. 8, pp. 1949–1954, 2024

  5. [4]

    Constrained nonlinear disturbance observer for robotic systems,

    J. W. Han, D. Park, and M. J. Kim, “Constrained nonlinear disturbance observer for robotic systems,” in 2024 IEEE International Conference on Robotics and Automation (ICRA) , pp. 3526–3532

  6. [5]

    Model-free friction observers for flexible joint robots with torque measurements,

    M. J. Kim, F. Beck, C. Ott, and A. Albu-Sch ¨affer, “Model-free friction observers for flexible joint robots with torque measurements,” IEEE Transactions on Robotics , vol. 35, no. 6, pp. 1508–1515, 2019

  7. [6]

    A passivity-based nonlinear admittance control with application to powered upper-limb control under unknown en- vironmental interactions,

    M. J. Kim, W. Lee, J. Y . Choi, G. Chung, K.-L. Han, I. S. Choi, C. Ott, and W. K. Chung, “A passivity-based nonlinear admittance control with application to powered upper-limb control under unknown en- vironmental interactions,” IEEE/ASME Transactions on Mechatronics, vol. 24, no. 4, pp. 1473–1484, 2019

  8. [7]

    A memory-based SO(3) parameterization: Theory and application to 6D impedance control with radially unbounded potential function,

    J. Jeong, H. Mishra, C. Ott, and M. J. Kim, “A memory-based SO(3) parameterization: Theory and application to 6D impedance control with radially unbounded potential function,” in IEEE International Conference on Robotics and Automation (ICRA) , pp. 8338–8344

Show all 39 references
  1. [8]

    Design of a momentum- based disturbance observer for rigid and flexible joint robots,

    M. J. Kim, Y . J. Park, and W. K. Chung, “Design of a momentum- based disturbance observer for rigid and flexible joint robots,” Intelli- gent Service Robotics , vol. 8, pp. 57–65, 2015

  2. [9]

    Collision detection for industrial collaborative robots: A deep learning approach,

    Y . J. Heo, D. Kim, W. Lee, H. Kim, J. Park, and W. K. Chung, “Collision detection for industrial collaborative robots: A deep learning approach,” IEEE Robotics and Automation Letters , vol. 4, no. 2, pp. 740–746, 2019

  3. [10]

    Transferable collision detection learning for collaborative manipulator using versatile modularized neural net- work,

    D. Kim, D. Lim, and J. Park, “Transferable collision detection learning for collaborative manipulator using versatile modularized neural net- work,” IEEE Transactions on Robotics, vol. 38, no. 4, pp. 2426–2445, 2022

  4. [11]

    Collision detection for robot manipulators using unsupervised anomaly detection algorithms,

    K. M. Park, Y . Park, S. Yoon, and F. C. Park, “Collision detection for robot manipulators using unsupervised anomaly detection algorithms,” IEEE/ASME Transactions on Mechatronics , vol. 27, no. 5, pp. 2841– 2851, 2022

  5. [13]

    Armcl: Arm contact point localization via monte carlo localization,

    A. Zwiener, R. Hanten, C. Schulz, and A. Zell, “Armcl: Arm contact point localization via monte carlo localization,” in 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pp. 7105–7111

  6. [14]

    Collision detection and isolation on a robot using joint torque sensing,

    J. Bimbo, C. Pacchierotti, N. G. Tsagarakis, and D. Prattichizzo, “Collision detection and isolation on a robot using joint torque sensing,” in 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pp. 7604–7609

  7. [15]

    Multi-scenario contacts handling for collaborative robots applica- tions,

    D. Popov, S. Mikhel, R. Yagfarov, A. Klimchik, and A. Pashkevich, “Multi-scenario contacts handling for collaborative robots applica- tions,” in 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pp. 2985–2992

  8. [16]

    Localizing external contact using pro- prioceptive sensors: The contact particle filter,

    L. Manuelli and R. Tedrake, “Localizing external contact using pro- prioceptive sensors: The contact particle filter,” in 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pp. 5062–5069

  9. [17]

    Easing reliance on collision-free planning with contact-aware control,

    T. Pang and R. Tedrake, “Easing reliance on collision-free planning with contact-aware control,” in 2022 International Conference on Robotics and Automation (ICRA) , pp. 8375–8381

  10. [18]

    Online multi-contact feedback model predictive control for interactive robotic tasks,

    S. W. Han, M. Iskandar, J. Lee, and M. J. Kim, “Online multi-contact feedback model predictive control for interactive robotic tasks,” in 2024 IEEE International Conference on Robotics and Automation (ICRA), pp. 11 556–11 562

  11. [19]

    Model predictive control for fast reaching in clutter,

    M. D. Killpack, A. Kapusta, and C. C. Kemp, “Model predictive control for fast reaching in clutter,” Autonomous Robots, vol. 40, pp. 537–560, 2016

  12. [20]

    Contact point localization for articulated manipulators with proprioceptive sensors and machine learning,

    A. Zwiener, C. Geckeler, and A. Zell, “Contact point localization for articulated manipulators with proprioceptive sensors and machine learning,” in 2018 IEEE International Conference on Robotics and Automation (ICRA), pp. 323–329

  13. [21]

    Transfer learning for collision localization in collaborative robotics,

    D. Popov and A. Klimchik, “Transfer learning for collision localization in collaborative robotics,” in Proceedings of the 3rd International Conference on Applications of Intelligent Systems , 2020, pp. 1–7

  14. [22]

    Identifying external con- tacts from joint torque measurements on serial robotic arms and its limitations,

    T. Pang, J. Umenberger, and R. Tedrake, “Identifying external con- tacts from joint torque measurements on serial robotic arms and its limitations,” in 2021 IEEE International Conference on Robotics and Automation (ICRA), pp. 6476–6482

  15. [23]

    Real-time estimation of multiple potential contact locations and forces,

    D. Popov, A. Klimchik, and A. Pashkevich, “Real-time estimation of multiple potential contact locations and forces,” IEEE Robotics and Automation Letters, vol. 6, no. 4, pp. 7025–7032, 2021

  16. [24]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,” Advances in neural information processing systems , vol. 33, pp. 6840–6851, 2020

  17. [25]

    Score-based generative modeling through stochastic differential equations,

    Y . Song, J. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, and B. Poole, “Score-based generative modeling through stochastic differential equations,” arXiv preprint arXiv:2011.13456 , 2020

  18. [26]

    Understanding diffusion models: A unified perspective,

    C. Luo, “Understanding diffusion models: A unified perspective,” arXiv preprint arXiv:2208.11970 , 2022

  19. [27]

    Collision detection and safe reaction with the dlr-iii lightweight manipulator arm,

    A. De Luca, A. Albu-Schaffer, S. Haddadin, and G. Hirzinger, “Collision detection and safe reaction with the dlr-iii lightweight manipulator arm,” in 2006 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pp. 1623–1630

  20. [28]

    Combining real and virtual sensors for measuring interaction forces and moments acting on a robot,

    G. Buondonno and A. De Luca, “Combining real and virtual sensors for measuring interaction forces and moments acting on a robot,” in 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 794–800

  21. [29]

    Collision detection, identification, and localization on the dlr sara robot with sensing redundancy,

    M. Iskandar, O. Eiberger, A. Albu-Sch ¨affer, A. De Luca, and A. Diet- rich, “Collision detection, identification, and localization on the dlr sara robot with sensing redundancy,” in 2021 IEEE International Conference on Robotics and Automation (ICRA) , pp. 3111–3117

  22. [30]

    Deep medial fields,

    D. Rebain, K. Li, V . Sitzmann, S. Yazdani, K. M. Yi, and A. Tagliasac- chi, “Deep medial fields,” arXiv preprint arXiv:2106.03804 , 2021

  23. [31]

    U-net: Convolutional networks for biomedical image segmentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” in Medical Image Computing and Computer-Assisted Intervention (MICCAI) , 2015

  24. [32]

    FiLM: Visual reasoning with a general conditioning layer,

    E. Perez, F. Strub, H. De Vries, V . Dumoulin, and A. Courville, “FiLM: Visual reasoning with a general conditioning layer,” in Pro- ceedings of the AAAI conference on artificial intelligence , vol. 32, no. 1, 2018

  25. [33]

    Fourier features let networks learn high frequency functions in low dimen- sional domains,

    M. Tancik, P. Srinivasan, B. Mildenhall, S. Fridovich-Keil, N. Ragha- van, U. Singhal, R. Ramamoorthi, J. Barron, and R. Ng, “Fourier features let networks learn high frequency functions in low dimen- sional domains,” Advances in neural information processing systems , vol. 33...

  26. [34]

    JAX: composable transformations of Python+NumPy programs,

    J. Bradbury, R. Frostig, P. Hawkins, M. J. Johnson, C. Leary, D. Maclaurin, G. Necula, A. Paszke, J. VanderPlas, S. Wanderman- Milne, and Q. Zhang, “JAX: composable transformations of Python+NumPy programs,” 2018. [Online]. Available: http://github. com/google/jax

  27. [35]

    qpSWIFT: A real-time sparse quadratic program solver for robotic applications,

    A. G. Pandala, Y . Ding, and H.-W. Park, “qpSWIFT: A real-time sparse quadratic program solver for robotic applications,” IEEE Robotics and Automation Letters, vol. 4, no. 4, pp. 3355–3362, 2019

  28. [36]

    MuJoCo: A physics engine for model-based control,

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

  29. [37]

    Denoising diffusion implicit models,

    J. Song, C. Meng, and S. Ermon, “Denoising diffusion implicit models,” arXiv preprint arXiv:2010.02502 , 2020

  30. [38]

    CDM: Contact Diffusion Model for Multi-Contact Point Localization (ICRA 2025),

    KAIST IRSL, “CDM: Contact Diffusion Model for Multi-Contact Point Localization (ICRA 2025),” (Jan. 31, 2025). Accessed: Jan. 31,

  31. [2025]

    Available: https://youtu.be/6kDZrnkFvp4

    [Online]. Available: https://youtu.be/6kDZrnkFvp4

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.