Pith. sign in

REVIEW 3 major objections 4 minor 26 references

Whole-body Multi-contact Motion Control for Humanoid Robots Based on Distributed Tactile Sensors

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Distributed tactile sensors on the limbs let a position-controlled humanoid robot balance during dynamic multi-contact motion that uses forearms, knees, and thighs, not just hands and feet.

desk verdict Solid engineering extension: intermediate-area tactile multi-contact control works on real hardware, but the tactile-specific region-update mechanism is only shown in simulation. read the letter →

arxiv 2505.19580 v1 pith:OJDSEAE5 submitted 2025-05-26 cs.RO

Masaki Murooka , Kensuke Fukumitsu , Marwan Hamze , Mitsuharu Morisawa , Hiroshi Kaminaga , Fumio Kanehiro , Eiichi Yoshida This is my paper · ORCID
classification cs.RO
keywords humanoidrobotswhole-bodymulti-contactmotiondistributedtactilesensorscontactregionestimationcentroidalmodelpredictivecontroldampingintermediatelimbcontactsreal-timebalance
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 sets out to show that a humanoid robot can keep its balance during dynamic whole-body motion while touching the world with body parts that carry no force sensors, such as forearms, knees, and thighs. The idea is to mount thin, deformable, sheet-shaped distributed tactile sensors on the limbs and feed their readings into the balance controller in two ways: as an online estimate of the true contact area, and as a measured contact wrench that drives compliant limb motion. In simulation, tactile feedback widens the range of environmental errors and external disturbances the robot survives; on the life-sized humanoid RHP Kaleido, the robot steps forward while leaning on a forearm and balances in a sitting posture supported only by thigh contact. This matters because confined spaces make intermediate-area contacts unavoidable, so robots that can lean on knees or elbows could operate where hand-and-foot-only controllers cannot.

What carries the argument

The load-bearing mechanism is an online contact-model update loop around the centroidal dynamics. In the resultant-wrench model $\bar w(\lambda,c)=\sum_i\sum_j\sum_k[\lambda_{i,j,k}\rho_{i,j,k};\,(p_{i,j}-c)\times\lambda_{i,j,k}\rho_{i,j,k}]$, the contact polygon vertices $p_{i,j}$ are normally fixed by the motion plan; here they are replaced on the fly by the smallest axis-aligned rectangle that encloses the tactile cells reporting contact. This corrected region flows into the model predictive centroidal controller and the wrench-distribution quadratic program, so the planned wrench respects what the real contact can support. On the limb side, the measured contact wrench $\mathbf w_a^i$ from the tactile cells drives the damping control law $K_d\Delta\dot r_c^i + K_s\Delta r_c^i = K_f(\mathbf w_a^i - \mathbf w_d^i)$, which shifts the contact area's compliance pose until the wrench error vanishes. The two loops together are what the paper credits for the stability gains: region updating prevents excessive moments on overestimated contacts, and wrench feedback lets the limb yield or press as disturbances demand.

What would settle it

On the physical robot, run the sitting-balance experiment using only tactile readings, with no proximity-sensor fallback, to estimate the thigh contact region: if RHP Kaleido falls whenever the tactile-based region estimate is active, then the claimed tactile-driven region update is not what stabilizes the real motion. A cheaper simulation check: inflate the true contact rectangle by 20 percent in the thigh-sitting scenario and verify that the robot falls exactly when the commanded moment exceeds the friction limit of the actual contact.

Watch

Extended reading notes

Core claim

The paper's central claim is that intermediate-area contacts can be brought into the same real-time balance-control loop as hand and foot contacts, provided the robot can feel them. The authors extend their earlier centroidal multi-contact controller so that distributed tactile sensors on the limbs supply two quantities: a measured contact wrench per contact area, computed from per-cell normal forces as $\mathbf w_a^i = \sum_s [f_\tau(\tau_{i,s})\nu_{i,s};\,(\xi_{i,s}-p_i)\times f_\tau(\tau_{i,s})\nu_{i,s}]$, and an updated contact polygon whose vertices are the smallest axis-aligned rectangle enclosing the active cells. The wrench drives a damping controller that adjusts each limb's compliance pose, while the polygon update keeps the centroidal model from commanding moments the true contact cannot generate. In simulation the combination widens the envelope of disturbances and environmental errors the robot survives: the tolerable wall-height error in elbow-contact walking extends from $-0.02$ m to $-0.03$ m on the lower bound, the survivable disturbance range in knee-contact standing grows from $[-90,40]$ N to $[-100,70]$ N, and thigh-contact sitting succeeds only when the contact region is updated from measurements. On hardware the same structure lets RHP Kaleido push about 160 N through its forearm while stepping forward and hold a sitting balance on a seat only 15 cm deep. The paper claims this is the first demonstration of dynamic motion with an intermediate-area contact on a position-controlled life-sized humanoid using real-time tactile feedback.

Load-bearing premise

The load-bearing premise is that the robot can always obtain a trustworthy, real-time estimate of the exact patch of limb surface that is in contact with the environment, since the paper shows that when the assumed contact area is larger than the real one the controller commands excessive moments and the robot falls, and on the real robot this estimate had to come from proximity sensors rather than from the tactile sensors themselves.

Editorial extensions

If this is right

  • Contact areas for balance control no longer need embedded 6-axis force/torque sensors; any limb surface covered by thin skin sensors can serve as a load-bearing contact.
  • Measurable robustness gains in simulation: the tolerable inclined-wall height error in elbow-contact walking extends from $-0.02$ m down to $-0.03$ m, and the survivable disturbance range in knee-contact standing widens from $[-90,40]$ N to $[-100,70]$ N.
  • The motions are dynamic, not quasi-static: the stepping experiment shifts the CoM about 0.2 m while exerting roughly 160 N through the forearm, and the sitting experiment balances on a seat only 15 cm deep.
  • An overestimated contact region is itself a failure mode: the paper shows the robot falls when the predefined region exceeds the true one, and that updating the region from sensor readings is what makes the thigh-sitting motions succeed.
  • The controller runs in real time on the robot's on-board computer, so the approach transfers from simulation to life-sized hardware.

Reading between the lines

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

  • The two-loop structure should transfer to other body surfaces, such as shins, upper arms, or chest, whenever a sensor patch is mounted and its intensity-to-force function $f_\tau$ is calibrated; the method is a general recipe for any-link balance support rather than a solution for three specific poses.
  • A testable consequence: deliberately inflating or shrinking the estimated contact rectangle should shift the disturbance limits in a predictable direction, which would make the region estimate usable as an online safety margin.
  • The authors' switch to proximity readings for region estimation in the real sitting experiment suggests the hardware bottleneck is the stability of tactile normal-force measurement at light contact, not the control law; a more reliable tactile-to-force calibration could remove that workaround.
  • The discrepancy between planned and measured contact regions is effectively a free sensor of environmental error, so it could trigger replanning of the contact sequence rather than only corrective feedback, an extension the paper lists as future work.
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

3 major / 4 minor

Summary. The paper extends an existing multi-contact controller for humanoid robots to contacts at intermediate limb areas (forearm, knee, thigh) by incorporating feedback from distributed tactile sensors. The centroidal MPC and wrench distribution modules are retained from prior work, with two extensions: contact-polygon vertices are updated online from tactile measurements, and limb damping control uses contact wrenches computed from tactile cell readings. MuJoCo simulations compare with and without tactile feedback for walking with elbow contact, standing with knee contact, and sitting with thigh contact; real experiments on RHP Kaleido demonstrate stepping with forearm support and sitting with thigh contacts. The authors claim that this is the first position-controlled life-sized humanoid to achieve dynamic motion with an intermediate-area contact through real-time tactile feedback.

Significance. If the claims hold, this is a useful advance for position-controlled humanoid robots: it relaxes the usual restriction to force/torque-instrumented hands and feet, and the authors release their control system and simulation environments as open-source code, which supports reproducibility. The controlled with/without tactile-feedback comparisons in simulation are a reasonable first-order test of the method, and the real forearm experiment gives direct evidence that tactile-based damping control can work on a life-sized platform. However, the hardware validation is partially decoupled from the paper's central tactile-sensing claim: the sitting experiment, which most exercises the contact-region update mechanism, uses proximity sensors for region estimation because the tactile readings were empirically less stable. Since overestimated contact regions are precisely the failure mode identified in simulation, the tactile-specific region-update mechanism is only validated in simulation. The reported per-motion tuning of damping gains also needs clarification before the robustness comparison can be fully interpreted.

major comments (3)
  1. [Section V-C] The sitting experiment, which is the main demonstration of contact-region update, used proximity sensors rather than tactile cells for region estimation: the paper states that 'the proximity sensors were used to estimate the contact region because we empirically found that the proximity sensor is more stable than the tactile sensor in estimating the contact region.' This is load-bearing because Section V-B shows that without contact-region update, the overestimated thigh region causes the robot to fall (Fig. 8). Therefore the real-robot result validates the controller with an online region update supplied by a different modality, not the tactile-cell-based region estimator that the paper's contribution (i) emphasizes. Please either provide hardware results using the tactile cells for region estimation, or explicitly reframe the contribution to say that the online region-update mechanism is validated in simulation with tactile sensing and on hardware with proximity sensing, and adjust the abstract/introduction wording accordingly.
  2. [Section IV-C] The contact-region estimator is defined as the smallest axis-aligned rectangle encompassing all cells where contact is detected, but the paper does not specify the contact-detection threshold, the handling of noise in the raw tactile intensity, or the behavior when active cells form disjoint clusters. A single bounding rectangle around disjoint clusters would overestimate the true contact region, reintroducing the excessive-moment failure mode shown in Fig. 8. Because the region update is central to the sitting result, please provide the threshold used for 'contact detected,' characterize the sensitivity of the estimated polygon to cell-level noise/occlusion, and either justify the single-rectangle assumption for the tested contact geometries or add a simulation validation with non-rectangular and disjoint contact patches.
  3. [Table II and Section V-A.2] The damping gains K_d at intermediate contact areas are reported as 'tuned for each motion in the range of 1000 to 100000 for translation and 100 to 1000 for rotation.' The robustness comparisons in Section V-B claim to isolate the effect of tactile feedback, but it is not stated whether the K_d, K_s, and K_f values were identical across the 'with' and 'without' tactile-feedback conditions, or whether K_d was re-tuned for each condition. Please state explicitly the gain values used in each comparison condition and explain how the comparison controls for gain tuning; otherwise, part of the reported robustness improvement could be attributed to the choice of K_d rather than to the tactile feedback itself.
minor comments (4)
  1. [Section V-B] The robustness results (wall-height error range, disturbance-force range) are reported as single intervals without specifying the sweep resolution or whether the simulator is deterministic; please state the increment used in the sweeps and whether any stochasticity (e.g., contact noise, joint noise) is present.
  2. [Section V-C, Fig. 12] The 'relatively large errors' in the forearm contact wrench are described qualitatively; please add a quantitative error measure (e.g., RMS or peak error over the contact phase) so that the reader can judge whether the tracking accuracy is acceptable for the claimed stable contact transition.
  3. [Section II] The e-skin cells include tactile, proximity, and acceleration sensing, but this is only explained in Section V-C; a brief mention in Section II would help the reader understand that the 'distributed tactile sensors' are multimodal, particularly because the sitting experiment uses the proximity mode.
  4. [Equation (5)] The term 'P_Alog' should be typeset as 'P_A log' for clarity; this is a minor formatting issue in the current preprint.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the controller derivation is self-contained and robustness claims are measured outcomes, not fitted inputs.

full rationale

The derivation is self-contained. The extension from extremity-only multi-contact control to intermediate-area contacts is built on explicit equations: resultant wrench (1), centroidal MPC (4), PD stabilization (5), wrench-distribution QP (7), damping control (9), and tactile-wrench conversion (11). The contact-polygon update is a stated approximation (smallest axis-aligned rectangle enclosing active cells, Section IV-C), not a hidden fit. The robustness comparisons in Section V-B are controlled experiments: the same planner and controller run with and without the tactile-feedback loop, and the measured ZMP, CoP, and fall thresholds are outcomes, not parameters fitted to produce the claim. The f_tau calibration in Section V-A1 is an external sensor-to-force mapping from force-gauge data, disclosed and not reused as evidence for the robustness result. Citations of the authors' prior work ([2], [15]) provide the base controller and are not invoked as a theorem that forces the present result. The real sitting experiment's use of proximity rather than tactile data for contact-region estimation (Section V-C) weakens the hardware validation of the tactile-specific region-update mechanism, but this is an experimental-limitation issue, not circularity: the controller equations do not reduce to their own inputs, and no fitted parameter is renamed as a prediction.

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

The method has no invented physical entities. It depends on several measurement and modeling assumptions: known contact schedule, normal-only tactile sensing, rectangular contact-region approximation, and the fidelity of MuJoCo. The control parameters (MPC weights, PD gains, damping gains, and the linear f_tau sensor map) are chosen by hand or fitted; K_d is explicitly tuned per motion, which limits the generality of the robustness claims.

free parameters (6)
  • MPC weight w_lambda = 5e-6
    Weight on control input in objective (4); chosen by hand.
  • MPC horizon N_h = 40
    Number of horizon steps in (4); chosen by hand.
  • MPC discretization period Delta_tau = 0.05 s
    Discretization period of the MPC horizon.
  • Stabilization gains P_L, D_L, P_A, D_A = diag(750,750,10000), diag(150,150,150), diag(750,750,750), diag(150,150,150)
    PD gains in Eq. (5); chosen by hand.
  • Damping control gains K_d, K_s, K_f = Contact phase: diag(10000,10000,10000,100,100,100), diag(0,0,0,0,0,2000), diag(1,1,1,1,1,0); K_d at intermediate areas…
    Gains in Eq. (9); tuned per motion in the specified ranges.
  • Tactile-to-force mapping f_tau = Not reported
    Linear fit between raw e-skin intensity and force gauge readings (Section V-A.1); coefficients are fitted per sensor and not stated.
assumptions (6)
  • domain assumption Contact sequence C_d (timing, location, area) is given
    Assumed in Section II; a global planner is cited as an alternative.
  • domain assumption Tactile cells measure only normal-direction response; tangential force and moment about the normal are zero in wrench reconstruction (Eq. 11)
    Section IV-C; damping feedback on those axes is disabled.
  • domain assumption Contact region is represented as the smallest axis-aligned rectangle enclosing all active cells
    Section IV-C; used for contact polygon update in (1).
  • domain assumption Robot is position-controlled with joint PD controllers
    Section II; damping control implemented through compliance pose passed to inverse kinematics.
  • domain assumption MuJoCo simulation with the tactile plugin faithfully models the robot and sensors
    Used for all simulation experiments in Section V-B.
  • domain assumption Feet-only ZMP within the support region implies the robot will not fall
    Stated in Section V-B footnote; an approximate stability criterion under non-coplanar contacts.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Whole-body Multi-contact Motion Control for Humanoid Robots Based on Distributed Tactile Sensors." pith.science (2026). https://pith.science/paper/OJDSEAE5

@misc{pith2026250519580,
  author       = {Pith},
  title        = {Pith review of: Whole-body Multi-contact Motion Control for Humanoid Robots Based on Distributed Tactile Sensors},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OJDSEAE5}},
  note         = {Machine review of arXiv:2505.19580}
}
read the original abstract

To enable humanoid robots to work robustly in confined environments, multi-contact motion that makes contacts not only at extremities, such as hands and feet, but also at intermediate areas of the limbs, such as knees and elbows, is essential. We develop a method to realize such whole-body multi-contact motion involving contacts at intermediate areas by a humanoid robot. Deformable sheet-shaped distributed tactile sensors are mounted on the surface of the robot's limbs to measure the contact force without significantly changing the robot body shape. The multi-contact motion controller developed earlier, which is dedicated to contact at extremities, is extended to handle contact at intermediate areas, and the robot motion is stabilized by feedback control using not only force/torque sensors but also distributed tactile sensors. Through verification on dynamics simulations, we show that the developed tactile feedback improves the stability of whole-body multi-contact motion against disturbances and environmental errors. Furthermore, the life-sized humanoid RHP Kaleido demonstrates whole-body multi-contact motions, such as stepping forward while supporting the body with forearm contact and balancing in a sitting posture with thigh contacts.

Figures

Figures reproduced from arXiv: 2505.19580 by the authors.

Figure 1
Figure 1. Control system for whole-body multi-contact motion in a humanoid robot. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Contact wrench representation. previous framework [2] by enhancing modules for centroidal control and limb control to adapt to whole-body contact based on tactile sensors in Sections III and IV, respectively. Then, in Section V, we present simulation and real-world experiments. In our control system, distributed tactile sensors are used to obtain the contact region for centroidal planning and wrench distribution, an… view at source ↗
Figure 3
Figure 3. Conversion of distributed tactile intensity to contact wrench. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: RHP Kaleido with distributed tactile sensors mounted on one forearm [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 6
Figure 6. Figure 6: Measurements when the robot is walking with elbow contact. [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]
Figure 7
Figure 7. Figure 7: Measurements when the robot is standing with knee contact. [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: Measurements when the robot is sitting with thigh contacts. [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]
Figure 10
Figure 10. Figure 10: Experiments on whole-body multi-contact motions by RHP Kaleido. [PITH_FULL_IMAGE:figures/full_fig_p007_10.png]
Figure 13
Figure 13. Figure 13: Force and CoP of the thighs during the motion in Fig. 10 (B). [PITH_FULL_IMAGE:figures/full_fig_p007_13.png]
Figure 12
Figure 12. Figure 12: Contact wrench of the forearm during the motion in Fig. 10 (A). [PITH_FULL_IMAGE:figures/full_fig_p007_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 22 canonical work pages

  1. [2]

    Centroidal trajectory generation and stabilization based on preview control for humanoid multi-contact motion,

    M. Murooka, M. Morisawa, and F. Kanehiro, “Centroidal trajectory generation and stabilization based on preview control for humanoid multi-contact motion,”IEEE Robotics and Automation Letters, vol. 7, no. 3, pp. 8225–8232, 2022

    work page 2022

  2. [1]

    Multi-contact vertical ladder climbing with an HRP-2 humanoid,

    J. Vaillant, A. Kheddar,et al., “Multi-contact vertical ladder climbing with an HRP-2 humanoid,”Autonomous Robots, vol. 40, pp. 561–580, 2016

    work page 2016

  3. [3]

    A comprehensive realization of robot skin: sensors, sensing, control, and applications,

    G. Chenget al., “A comprehensive realization of robot skin: sensors, sensing, control, and applications,”Proceedings of the IEEE, vol. 107, no. 10, pp. 2034–2051, 2019

    work page 2019

  4. [4]

    A flexible and robust large scale capacitive tactile system for robots,

    P. Maiolino, M. Maggiali, G. Cannata, G. Metta, and L. Natale, “A flexible and robust large scale capacitive tactile system for robots,”IEEE Sensors Journal, vol. 13, no. 10, pp. 3910–3917, 2013

    work page 2013

  5. [5]

    Accessed: 2023-10-1

    REEM-C. Accessed: 2023-10-1. [Online]. Available: https: //pal-robotics.com/robots/reem-c/

    work page 2023

  6. [6]

    iCub: The not-yet-finished story of building a robot child,

    L. Natale, C. Bartolozzi, D. Pucci, A. Wykowska, and G. Metta, “iCub: The not-yet-finished story of building a robot child,”Science Robotics, vol. 2, no. 13, p. eaaq1026, 2017

    work page 2017

  7. [7]

    Spatial calibra- tion of humanoid robot flexible tactile skin for human-robot interaction,

    S. Chefchaouni Moussaoui, R. Cisneros-Lim ´on,et al., “Spatial calibra- tion of humanoid robot flexible tactile skin for human-robot interaction,” Sensors, vol. 23, no. 9, 2023

    work page 2023

  8. [8]

    Whole-body active compliance control for humanoid robots with robot skin,

    E. Dean-Leon, J. R. Guadarrama-Olvera, F. Bergner, and G. Cheng, “Whole-body active compliance control for humanoid robots with robot skin,” inIEEE International Conference on Robotics and Automation, 2019, pp. 5404–5410

    work page 2019

Show all 26 references
  1. [9]

    Realizing whole-body tactile interactions with a self-organizing, multi-modal artificial skin on a humanoid robot,

    P. Mittendorfer, E. Yoshida, and G. Cheng, “Realizing whole-body tactile interactions with a self-organizing, multi-modal artificial skin on a humanoid robot,”Advanced Robotics, vol. 29, no. 1, pp. 51–67, 2015

    2015

  2. [10]

    Detecting grasp phases and adaption of object-hand interaction forces of a soft humanoid hand based on tactile feedback,

    P. Weiner, F. Hundhausen, R. Grimm, and T. Asfour, “Detecting grasp phases and adaption of object-hand interaction forces of a soft humanoid hand based on tactile feedback,” inIEEE/RSJ International Conference on Intelligent Robots and Systems, 2021, pp. 3979–3986

    2021

  3. [11]

    Towards effective tactile identification of textures using a hybrid touch approach,

    T. Taunyazov, H. F. Koh, Y . Wu, C. Cai, and H. Soh, “Towards effective tactile identification of textures using a hybrid touch approach,” inIEEE International Conference on Robotics and Automation, 2019, pp. 4269– 4275

    2019

  4. [12]

    Plantar tactile feedback for biped balance and locomotion on unknown terrain,

    J. Rogelio Guadarrama Olvera, E. D. Leon, F. Bergner, and G. Cheng, “Plantar tactile feedback for biped balance and locomotion on unknown terrain,”International Journal of Humanoid Robotics, vol. 17, no. 1, p. 1950036, 2020

    2020

  5. [13]

    Multi-contact motion planning and control,

    K. Bouyarmane, S. Caron, A. Escande, and A. Kheddar, “Multi-contact motion planning and control,” inHumanoid Robotics: A Reference. Springer, 2018, pp. 1763–1804

    2018

  6. [14]

    Robust vertical ladder climbing and transitioning between ladder and catwalk for humanoid robots,

    M. Kanazawa, S. Nozawa, Y . Kakiuchi, Y . Kanemoto, M. Kuroda, K. Okada, M. Inaba, and T. Yoshiike, “Robust vertical ladder climbing and transitioning between ladder and catwalk for humanoid robots,” in IEEE/RSJ International Conference on Intelligent Robots and Systems, 2015,...

    2015

  7. [15]

    Multi-contact stabilization of a humanoid robot for realizing dynamic contact transitions on non-coplanar sur- faces,

    M. Morisawa, M. Benallegue, R. Cisneros, I. Kumagai, A. Escande, K. Kaneko, and F. Kanehiro, “Multi-contact stabilization of a humanoid robot for realizing dynamic contact transitions on non-coplanar sur- faces,” inIEEE/RSJ International Conference on Intelligent Robots and Sy...

    2019

  8. [16]

    Toward whole-body loco- manipulation: Experimental results on multi-contact interaction with the Walk-Man robot,

    E. Farnioli, M. Gabiccini, and A. Bicchi, “Toward whole-body loco- manipulation: Experimental results on multi-contact interaction with the Walk-Man robot,” inIEEE/RSJ International Conference on Intelligent Robots and Systems, 2016, pp. 1372–1379

    2016

  9. [17]

    Online generation and control of quasi-static multi-contact motion by pwt ja- cobian matrix with contact wrench estimation and joint load reduction,

    N. Hiraoka, M. Murooka, S. Noda, K. Okada, and M. Inaba, “Online generation and control of quasi-static multi-contact motion by pwt ja- cobian matrix with contact wrench estimation and joint load reduction,” Advanced Robotics, vol. 35, no. 1, pp. 48–63, 2021

    2021

  10. [18]

    Multi-contact balancing of humanoid robots in confined spaces: Utilizing knee contacts,

    B. Henze, A. Dietrich, M. A. Roa, and C. Ott, “Multi-contact balancing of humanoid robots in confined spaces: Utilizing knee contacts,” in IEEE/RSJ International Conference on Intelligent Robots and Systems, 2017, pp. 697–704

    2017

  11. [19]

    Control-limited differential dynamic programming,

    Y . Tassa, N. Mansard, and E. Todorov, “Control-limited differential dynamic programming,” inIEEE International Conference on Robotics and Automation, 2014, pp. 1168–1175

    2014

  12. [20]

    Whole-body model predictive control for biped locomotion on a torque-controlled humanoid robot,

    E. Dantec, M. Naveau, P. Fernbach, N. Villa, G. Saurel, O. Stasse, M. Taix, and N. Mansard, “Whole-body model predictive control for biped locomotion on a torque-controlled humanoid robot,” inIEEE-RAS International Conference on Humanoid Robots, 2022, pp. 638–644

    2022

  13. [21]

    A unified MPC framework for whole-body dynamic locomotion and manipula- tion,

    J.-P. Sleiman, F. Farshidian, M. V . Minniti, and M. Hutter, “A unified MPC framework for whole-body dynamic locomotion and manipula- tion,”IEEE Robotics and Automation Letters, vol. 6, no. 3, pp. 4688– 4695, 2021

    2021

  14. [22]

    Biped walking stabilization based on linear inverted pendulum tracking,

    S. Kajitaet al., “Biped walking stabilization based on linear inverted pendulum tracking,” inIEEE/RSJ International Conference on Intelli- gent Robots and Systems, 2010, pp. 4489–4496. MUROOKAet al.: HUMANOID MULTI-CONTACT MOTION BASED ON TACTILE SENSORS 9

    2010

  15. [23]

    Development of life-sized humanoid robot platform with robustness for falling down, long time working and error occurrence,

    Y . Kakiuchi, M. Kamon, N. Shimomura, S. Yukizaki, N. Takasugi, S. Nozawa, K. Okada, and M. Inaba, “Development of life-sized humanoid robot platform with robustness for falling down, long time working and error occurrence,” inIEEE/RSJ International Conference on Intelligent R...

    2017

  16. [24]

    Accessed: 2023-10-1

    mc rtc. Accessed: 2023-10-1. [Online]. Available: https://jrl-umi3218. github.io/mc rtc/

    2023

  17. [25]

    MuJoCo: A physics engine for model-based control,

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

    2012

  18. [26]

    mc-mujoco: Simulating articulated robots with FSM controllers in MuJoCo,

    R. P. Singh, P. Gergondet, and F. Kanehiro, “mc-mujoco: Simulating articulated robots with FSM controllers in MuJoCo,” inIEEE/SICE International Symposium on System Integration, 2023, pp. 1–5

    2023

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.