Design of a Five-Fingered Hand with Full-Fingered Tactile Sensors Using Conductive Filaments and Its Application to Bending after Insertion Motion

Kazuhiro Miyama; Kei Okada; Kento Kawaharazuka; Masayuki Inaba; Naoya Yamaguchi; Shun Hasegawa

REVIEW 3 major objections 6 minor 13 references

Design of a Five-Fingered Hand with Full-Fingered Tactile Sensors Using Conductive Filaments and Its Application to Bending after Insertion Motion

T0 review · 3 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A five-fingered robotic hand with a 3D-printed skin containing conductive 'nerve lines' estimates contact position along the entire finger and uses this feedback to insert its fingers into scissors, slide the tool to the base, and operate…

desk verdict A genuinely new low-cost full-finger tactile skin with a clever conductive-filament design, but the position-estimation claim is only calibrated on bare nerve lines and the system-level demonstration is threshold-level, so the paper is a promising prototype rather than a validated sensor. read the letter →

arxiv 2412.00732 v1 pith:JNSU7JNZ submitted 2024-12-01 cs.RO

Kazuhiro Miyama , Shun Hasegawa , Kento Kawaharazuka , Naoya Yamaguchi , Kei Okada , Masayuki Inaba This is my paper · ORCID

classification cs.RO

keywords five-fingeredhandnerveinclusionflexibleepidermisconductivefilamentsensorcontactpointestimationbendingafterinsertionmotionscissorsmanipulationVoronoistructurewire-driven

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

This paper claims that a full-finger tactile sense can be built into a robot hand by 3D printing a flexible skin with embedded conductive-filament 'nerve lines,' and that this sense is enough to let a thin-fingered hand manipulate tools that require inserting the fingers and then bending them. The authors construct a five-fingered wire-driven hand with human-like finger thickness, cover it with this nerve-inclusion epidermis, and use the contact-position estimate from resistance changes to guide a scissors-insertion task. The approach matters because most tactile sensors cover only the fingertip or palm, while tool use such as opening scissors needs contact feedback along the whole finger, including the joints and both the palm and back sides. The paper demonstrates the 'bending after insertion motion': insert the finger into the scissors ring, slide the scissors to the base of the finger, bend the finger to fix the tool, and operate the scissors.

What carries the argument

The nerve-inclusion flexible epidermis: a 3D-printed Voronoi-patterned flexible skin that covers the hand and holds conductive-filament nerve lines. Each nerve line consists of an insulated flat part along the skeleton and an insulated string-like part with 5 mm spikes pulled through the mesh holes, so contact pushes the string onto the flat part, closing the circuit; the resistance then encodes the distance of the contact from the wire junction. The piecewise-linear calibration equation (Eq. 1) maps the measured voltage to a contact-position ratio $p$ using the three reference voltages $V_{\max}$, $V_{\min}$, and $V_{\mathrm{mid}}$. The Voronoi mesh, with low density at joints and high density where strength is needed, provides the flexibility and exposes the spikes, while the hand's wire-driven seven-actuator design keeps the fingers thin enough for insertion.

What would settle it

Bend the index finger through its full range with no external contact and record the estimated contact ratio $p$ from Equation (1); if $p$ falls below the grasp-detection threshold of 90 from joint motion alone, or if a known single contact at a fixed distance is reported at a different position when the finger is bent versus straight, the core monotonic-resistance assumption fails in the exact regime the tool-use task requires.

Watch

Extended reading notes

Core claim

At the core of the paper is the claim that a two-part, insulated nerve line made from conductive TPU filament, embedded in a Voronoi-patterned flexible 3D-printed epidermis, can locate a contact anywhere along a robotic finger by a single resistance reading. The nerve line has a flat part laid against the skeleton and a string-like part with spikes at 5 mm intervals that protrude through the skin's mesh; when contact presses the string onto the flat part, the circuit closes and the resistance grows with the distance of the contact from the wire junction. The method calibrates three voltages—steady state $V_{\max}$, tip contact $V_{\min}$, and an intermediate reference $V_{\mathrm{mid}}$—and uses the piecewise-linear Equation (1) to output a contact-position ratio $p$ between 0 and 100. Experiments on a wire-driven five-fingered hand with joint widths of 9–14 mm show that $p$ tracks a sliding contact linearly, and the authors use thresholds on $p$ to detect grasp success and to detect that scissors have been pushed to the base of the finger before bending, demonstrating the 'bending after insertion' motion.

Load-bearing premise

The position estimate works only if the resistance of a nerve line changes monotonically and repeatably with the distance of a single contact from the wire junction, and if the piecewise-linear calibration between $V_{\max}$, $V_{\min}$, and $V_{\mathrm{mid}}$ remains valid under the variable forces, bending, and multi-point contacts that occur during real tool manipulation.

Editorial extensions

If this is right

The same fully 3D-printed sensing skin can be fitted to any thin-fingered hand, adding contact feedback on both the palmar and dorsal sides, including the joints, without changing the hand's structure.
The contact-position ratio $p$ can act as a grasp-success criterion: the scissors-lifting experiment uses a threshold of $p=90$ to decide whether the insertion was successful.
The same $p$ value serves as a position cue during regrasping, with a threshold of $p=50$ used to detect that the scissors have been pushed to the base of the finger before bending.
Because the sensor is removable and 3D-printed, it can be reproduced or reconfigured quickly for different hand sizes or tool shapes.

Reading between the lines

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

The paper only tests single-contact localization; time-multiplexing or frequency-tagging multiple nerve lines could let the same skin resolve two simultaneous contacts, which the current method explicitly cannot do.
Since the resistance also varies with how hard the string part is pressed through the mesh, a recalibrated version of the sensor could estimate contact force in addition to position, enabling force-controlled grasping with the same skin.
The Voronoi mesh density is chosen manually per joint; an automated design step over mesh density and spike spacing could adapt the nerve-inclusion principle to other hand geometries and tool types.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Colleague, here is my read on Miyama et al. The genuinely new bit is the full-finger coverage: a nerve line made of conductive TPU, split into insulated flat and spiked string parts, embedded in a Voronoi-patterned flexible 3D-printed epidermis, covering dorsal and palmar sides of index and middle fingers including the joints. That combination is not in the prior work they cite (mostly fingertip or pad sensors, and the jointless five-fingered tactile skin in [11]). The hardware story is coherent: the flat part lies against the skeleton, the string part pokes through the Voronoi holes, and contact closes the circuit; resistance gives position; four nerves on two fingers. The "bending after insertion" task, with a five-fingered wire-driven hand whose finger joints are sized near human dimensions, is a sensible application and the scissors and pen demos are plausible.

Now the soft spots, and they are real. The calibration in Section II-D is done on bare nerve lines, pressed at 5 mm intervals, with and without spikes. It is not done on the assembled epidermis, on the 3D-printed finger, or with the deformation path through the epidermis holes. The stress-test note is right: Eq. (1) maps voltage to position using Vmax, Vmid, Vmin from that bare-line calibration, and nothing in the paper shows that the map survives contact force, joint angle, or the specific spike geometry engaged through the skin. Larger variance with spikes is acknowledged, but no error bars, per-trial data, or statistics are given. The scissors operation uses only thresholds (p=90, p=50), so it demonstrates contact-event detection, not continuous position estimation under task conditions. The authors do state some limitations: the nonlinearity, the inability to localize simultaneous contacts, and the need for calibration per nerve line. That is honest, but it does not close the validation gap.

The finger design section is light: the wire-to-angle relation x=rθ is trivial, and the optimal servo displacement is found by manual marking and pulling. That is fine for a prototype but not a design method.

Who is this for? People building low-cost soft tactile skins for wire-driven hands. It deserves a serious referee, not a desk rejection, because the hardware idea is novel and the paper is clearly written. But a referee should insist on an integrated calibration, repeated trials with quantitative error, and at least one task-condition comparison between bare-line and epidermis-covered behavior. As is, it is a promising prototype paper, not a validated sensing system.

Referee Report

3 major / 6 minor

Summary. The paper presents a nerve inclusion flexible epidermis, a 3D-printed compliant skin with embedded conductive-filament 'nerve lines,' intended to detect contact positions on both sides of a finger including the joints. The sensor is based on measuring resistance changes when a string-like, spiked conductive part contacts a flat conductive part through holes in a Voronoi-patterned skin. The authors also describe a five-fingered wire-driven hand with thin, human-like fingers and demonstrate a tool-use task: inserting fingers into scissors and bending them to operate the scissors, using the sensor output as contact feedback. The central claim is that the proposed method can estimate the contact point over the full finger and that this information enables the bending-after-insertion motion.

Significance. If the positional estimation claim holds in the integrated system, the paper offers a low-cost, easily fabricated full-finger tactile skin and a robotic hand with slender fingers suitable for insertion into human tools. The combination of a Voronoi-patterned epidermis with conductive-filament nerve lines is a novel sensor concept, and the integration into a five-fingered hand with a demonstrated tool-use task is a valuable system contribution. The paper includes a concrete sensor construction and calibration procedure. However, the current evidence does not substantiate the claimed full-finger contact-position estimation under realistic task conditions: the calibration experiment is performed only on a bare nerve line, and the tool-use experiments use the sensor as a binary threshold rather than as a continuous position estimator. This gap between the stated claim and the demonstrated evidence is the main weakness.

major comments (3)

[Section II-D, Fig. 8] The position estimation experiment is performed only on a bare nerve line (with and without spikes), not on the nerve-inclusion epidermis attached to the finger. The actual sensor geometry on the robot includes the Voronoi mesh holes, curved finger surfaces, and joint bending, all of which alter how the string-like part contacts the flat part and thus change the resistance mapping. Without calibrating and validating Eq. (1) on the assembled finger with ground-truth contact positions at various normal forces and joint angles, the claim that the sensor estimates contact position over the full finger including joints is not established.
[Section II-D, Fig. 8] The reported validation results give only means and variances, without per-trial data, confidence intervals, or error metrics such as mean absolute error or root mean square error. The text itself notes that the variance is larger in the spiked configuration and that the estimate depends on which spike makes contact. Because the robot hand uses the spiked configuration, the paper should quantify the estimation error and repeatability for that configuration and discuss how the thresholds p=90 and p=50 relate to the position uncertainty.
[Sections IV-A and IV-B] The tool-use experiments use the sensor output only as a binary threshold (p < 90 to detect grasping, p < 50 to detect that the scissors reached the finger base). This demonstrates contact-event detection, but it does not demonstrate continuous full-finger contact-position estimation during the task, which is a central claimed capability. Furthermore, the experiments appear to be single demonstrations without repeated trials, so the robustness and variability of the method under task conditions are unquantified. Repeated trials and reported sensor traces (p values over time) are needed to support the claim that the sensor provides position information useful for tool use.

minor comments (6)

[Section II-B] The phrase 'The strings-like part have spikes' should be 'The string-like part has spikes,' and the repeated statement that 'both parts are insulated' could be rephrased for clarity.
[Section II-D] There is a typo 'estimated valuee p' in the description of the experiment; also, Fig. 8's caption should specify whether the error bars denote standard deviation or another measure.
[Section IV-B] The text says 'mother finger' where 'thumb' is intended; the caption of Fig. 14 also contains the typo 'Liftting up.'
[Section IV] The robot system 'HIRONX' is mentioned without definition; the authors should briefly explain what it is (e.g., a humanoid torso) and its role in the experiments.
[Section III, Eq. (2)] The relationship x = rθ is stated for wire displacement and rotation angle, but the text then says the optimal servo displacement was determined manually. Clarify for which actuators Eq. (2) is actually used and whether it is a design equation or a controller mapping.
[Abstract and Conclusion] The abstract and conclusion claim that contact position estimation is achieved and that the effectiveness is confirmed, but the experiments in Section IV only demonstrate threshold-based contact detection. The wording should be tempered to match the evidence, or additional evidence should be provided.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the contact-position mapping is a calibrated interpolation, and the tool-use demonstrations do not repackage fitted inputs as predictions.

full rationale

The paper's central sensing claim rests on a calibration-plus-interpolation procedure rather than on a derivation that assumes its own conclusion. Equation (1) defines the contact ratio p using three measured reference voltages, Vmax, Vmid, and Vmin, which are obtained by direct calibration at known sensor locations (Section II-D). The validation experiment then presses the nerve line at 5 mm intervals and compares the resulting p values against the known positions; the intermediate points are not used to fit the mapping, so this is a genuine interpolation test, not a circular prediction. The tool-use experiments use hand-set thresholds (p = 90 and p = 50) to trigger control decisions, but they do not claim to have predicted those thresholds from the sensor model, so no fitted input is renamed as a prediction. The only overlap with the authors' prior work is the background citation [10] (Hirose, Kakiuchi, Okada, and Inaba) for earlier soft wire-driven finger and epidermis research; that citation is not load-bearing for the proposed nerve-inclusion epidermis, and the Voronoi structure is attributed to external prior work [12]. Concerns about whether the bare-nerve calibration transfers to the assembled epidermis, force variation, bending, or multi-point contact are validity or robustness questions, not circularity. The derivation chain is therefore self-contained, and no circular step can be exhibited from the paper's own equations or citations.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

No theoretical entities are introduced. The nerve inclusion flexible epidermis is a physical hardware module whose components are described. The central claim rests on calibrated sensor parameters and domain assumptions about the monotonicity and stability of the conductive-filament resistance mapping, rather than on new postulated physical objects.

free parameters (6)

Vmax calibration voltage = measured per nerve line, value not reported
Steady-state sensor voltage used as the upper bound in Eq. 1 for contact point ratio estimation.
Vmid calibration voltage = measured per nerve line, value not reported
Voltage when the nerve line tip conducts; anchors the 20% fingertip segment in Eq. 1.
Vmin calibration voltage = measured per nerve line, value not reported
Minimum sensor voltage at the far calibration point; anchors the 80% segment in Eq. 1.
Contact detection threshold = p = 90
Hand-set threshold used in the scissor grasping experiment to decide that contact has occurred.
Contact position threshold for scissor base = p = 50
Hand-set threshold used to decide that the scissors have moved to the base of the finger.
Fingertip segment fraction = 20%
Equation 1 allocates 20% of the contact point ratio range to the fingertip segment; this split is a design choice rather than a derived quantity.

assumptions (4)

domain assumption The resistance of the conductive filament nerve line changes monotonically with the distance of a single contact point from the wire junction.
All position estimates in Eq. 1 rely on this monotonic relationship; the paper provides calibration evidence but no physical derivation, as described in Section II-B.
domain assumption The insulation between the flat and string-like parts of the nerve line remains intact during bending and repeated contact, so the no-contact state remains distinguishable.
The sensing scheme depends on the two parts being insulated until pressed together; Section II-B explains why the insulation was chosen.
domain assumption Contact at the fingertip region can be separated from contact elsewhere by a fixed 20% piecewise-linear segment of the calibration curve.
Equation 1 treats the range Vmid < v < Vmax as the fingertip 20%, a modeling choice not derived from a physical model.
domain assumption Tool manipulation can be treated as effectively single-contact, so the sensor's inability to localize two or more simultaneous contacts is not task-limiting.
The authors state this limitation in Section II-C and argue tools are stable when placed behind the finger; this is asserted, not measured.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Design of a Five-Fingered Hand with Full-Fingered Tactile Sensors Using Conductive Filaments and Its Application to Bending after Insertion Motion." pith.science (2026). https://pith.science/paper/JNSU7JNZ

@misc{pith2026241200732,
  author       = {Pith},
  title        = {Pith review of: Design of a Five-Fingered Hand with Full-Fingered Tactile Sensors Using Conductive Filaments and Its Application to Bending after Insertion Motion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JNSU7JNZ}},
  note         = {Machine review of arXiv:2412.00732}
}

read the original abstract

The purpose of this study is to construct a contact point estimation system for the both side of a finger, and to realize a motion of bending the finger after inserting the finger into a tool (hereinafter referred to as the bending after insertion motion). In order to know the contact points of the full finger including the joints, we propose to fabricate a nerve inclusion flexible epidermis by combining a flexible epidermis and a nerve line made of conductive filaments, and estimate the contact position from the change of resistance of the nerve line. A nerve inclusion flexible epidermis attached to a thin fingered robotic hand was combined with a twin-armed robot and tool use experiments were conducted. The contact information can be used for tool use, confirming the effectiveness of the proposed method.

Figures

Figures reproduced from arXiv: 2412.00732 by the authors.

**Figure 1.** system config of the proposed method degree of freedom to grasp and use scissors. It is expected that the ability to perform finger insertion into tools and environments in addition to common grasping tasks will greatly aid in the imitation of human tasks. Based on the above, the following two elements are considered to be necessary for the post-insertion bending operation. • Tactile sensation of the both side of th… view at source ↗

**Figure 2.** Differences in mesh density settings by part. Low density is used for joints that move a lot, and high density is used for parts that require strength. B. Neural Line Structure and Contact Point Estimation Method The nerve line was made with a conductive TPU filament called Conductive Filaflex(produced by Recreus). The nerve line can be roughly divided into a flat part (A in [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 3.** Overview of a nerve line. The two parts are insulated and their [PITH_FULL_IMAGE:figures/full_fig_p002_3.png] view at source ↗

Figures from the paper (10 more)

**Figure 4.** Figure 4: A cross section of the finger. The nerve line is arranged across the [PITH_FULL_IMAGE:figures/full_fig_p003_4.png]

**Figure 6.** Figure 6: The schematic of a sensor cuircuit. This is consists of nerve line [PITH_FULL_IMAGE:figures/full_fig_p003_6.png]

**Figure 7.** Figure 7: In this experiment, the effective sensor length was set to 80[mm], and the sensor performance was evaluated by comparing the change in the estimated valuee p when the nerve line is contacted at intervals of 5[mm]=6.25[%]. In addition, to verify the change in performanc…

**Figure 9.** Figure 9: Overall view of the developed device and the functions of the seven [PITH_FULL_IMAGE:figures/full_fig_p004_9.png]

**Figure 10.** Figure 10: Example of utilizing the internal rotation mechanism. Combined [PITH_FULL_IMAGE:figures/full_fig_p005_10.png]

**Figure 11.** Figure 11: Contact point ratio P when grasping the scissors [PITH_FULL_IMAGE:figures/full_fig_p005_11.png]

**Figure 12.** Figure 12: Figure of scissors grasping experiment using nerve inclusion [PITH_FULL_IMAGE:figures/full_fig_p005_12.png]

**Figure 13.** Figure 13: Contact point ratio P when the scissors are regrasped and used. [PITH_FULL_IMAGE:figures/full_fig_p006_13.png]

**Figure 14.** Figure 14: Figure of scissors operating experiment using nerve inclusion [PITH_FULL_IMAGE:figures/full_fig_p006_14.png]

**Figure 15.** Figure 15: The scissors are fixed in place by inserting and bending the fingers, [PITH_FULL_IMAGE:figures/full_fig_p006_15.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

13 extracted references · 13 canonical work pages

[11]

Object recognition through active sensing using a multi-fingered robot hand with 3d tactile sensors,

S. Funabashi, S. Morikuni, A. Geier, A. Schmitz, S. Ogasa, T. P. Torno, S. Somlor, and S. Sugano, “Object recognition through active sensing using a multi-fingered robot hand with 3d tactile sensors,” in 2018 IEEE /RSJ International Conference on Intelligent Robots and Systems, 2018, pp. 2589–2595

work page 2018
[1]

Dlr-hand ii: next generation of a dextrous robot hand,

J. Butterfass, M. Grebenstein, H. Liu, and G. Hirzinger, “Dlr-hand ii: next generation of a dextrous robot hand,” in 2001 IEEE International Conference on Robotics and Automation , vol. 1, 2001, pp. 109–114

work page 2001
[2]

A parallel gripper with capability to use various tools,

T. Akaike, Y . Tsumaki, R. Tadakuma, and M. Yamano, “A parallel gripper with capability to use various tools,” in 2012 SICE Annual Conference, 2012, pp. 2139–2143

work page 2012
[3]

Development of humanoid hand with cover integrated link mechanism for daily life work,

N. Fukaya and Y . Ogasawara, “Development of humanoid hand with cover integrated link mechanism for daily life work,” in 2017 IEEE 6th Global Conference on Consumer Electronics , 2017, pp. 1–4

work page 2017
[4]

High-speed high-precision proximity sensor for detection of tilt, distance, and contact,

K. Koyama, M. Shimojo, T. Senoo, and M. Ishikawa, “High-speed high-precision proximity sensor for detection of tilt, distance, and contact,” IEEE Robotics and Automation Letters , vol. 3, no. 4, pp. 3224–3231, Oct. 2018

work page 2018
[5]

Electric field servoing for robotic ma- nipulation,

R. Wistort and J. R. Smith, “Electric field servoing for robotic ma- nipulation,” in 2008 IEEE/RSJ International Conference on Intelligent Robots and Systems , Sept. 2008, pp. 494–499

work page 2008
[6]

6d proximity servoing for preshaping and haptic exploration using capacitive tactile proximity sensors,

S. Escaida Navarro, M. Schonert, B. Hein, and H. W ¨orn, “6d proximity servoing for preshaping and haptic exploration using capacitive tactile proximity sensors,” in 2014 IEEE /RSJ International Conference on Intelligent Robots and Systems , Sept. 2014, pp. 7–14

work page 2014
[7]

Seashell e ffect pretouch sensing for robotic grasping,

L.-T. Jiang and J. R. Smith, “Seashell e ffect pretouch sensing for robotic grasping,” in 2012 IEEE International Conference on Robotics and Automation , May 2012, pp. 2851–2858

work page 2012

Show all 13 references

[8]

Stretchable capacitive tactile skin on humanoid robot fingers-first experiments and results,

A. P. Gerratt, N. Sommer, S. P. Lacour, and A. Billard, “Stretchable capacitive tactile skin on humanoid robot fingers-first experiments and results,” in 2014 IEEE-RAS International Conference on Humanoid Robots, 2014, pp. 238–245

2014
[9]

A robot gripper with variable stiffness actuation for enhancing collision safety,

A. H. Memar and E. T. Esfahani, “A robot gripper with variable stiffness actuation for enhancing collision safety,” IEEE Transactions on Industrial Electronics , vol. 67, no. 8, pp. 6607–6616, 2020

2020
[10]

Design of soft flex- ible wire-driven finger mechanism for contact pressure distribution,

T. Hirose, Y . Kakiuchi, K. Okada, and M. Inaba, “Design of soft flex- ible wire-driven finger mechanism for contact pressure distribution,” in 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2019, pp. 4699–4705

2019
[12]

3d-architected soft machines with topologically encoded motion,

D. Goswami, S. Liu, A. Pal, L. G. Silva, and R. V . Martinez, “3d-architected soft machines with topologically encoded motion,” Advanced Functional Materials , vol. 29, no. 24, p. 1808713, 2019

2019
[13]

Developing a hand sizing system for a hand exoskeleton device based on the kansei engineering method,

C.-C. Kuo, H.-Y . Kung, H.-C. Wu, and M.-J. Wang, “Developing a hand sizing system for a hand exoskeleton device based on the kansei engineering method,” Journal of Ambient Intelligence and Humanized Computing, pp. 1–13, 2020

2020

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [11]

Object recognition through active sensing using a multi-fingered robot hand with 3d tactile sensors,

S. Funabashi, S. Morikuni, A. Geier, A. Schmitz, S. Ogasa, T. P. Torno, S. Somlor, and S. Sugano, “Object recognition through active sensing using a multi-fingered robot hand with 3d tactile sensors,” in 2018 IEEE /RSJ International Conference on Intelligent Robots and Systems, 2018, pp. 2589–2595

work page 2018

[2] [1]

Dlr-hand ii: next generation of a dextrous robot hand,

J. Butterfass, M. Grebenstein, H. Liu, and G. Hirzinger, “Dlr-hand ii: next generation of a dextrous robot hand,” in 2001 IEEE International Conference on Robotics and Automation , vol. 1, 2001, pp. 109–114

work page 2001

[3] [2]

A parallel gripper with capability to use various tools,

T. Akaike, Y . Tsumaki, R. Tadakuma, and M. Yamano, “A parallel gripper with capability to use various tools,” in 2012 SICE Annual Conference, 2012, pp. 2139–2143

work page 2012

[4] [3]

Development of humanoid hand with cover integrated link mechanism for daily life work,

N. Fukaya and Y . Ogasawara, “Development of humanoid hand with cover integrated link mechanism for daily life work,” in 2017 IEEE 6th Global Conference on Consumer Electronics , 2017, pp. 1–4

work page 2017

[5] [4]

High-speed high-precision proximity sensor for detection of tilt, distance, and contact,

K. Koyama, M. Shimojo, T. Senoo, and M. Ishikawa, “High-speed high-precision proximity sensor for detection of tilt, distance, and contact,” IEEE Robotics and Automation Letters , vol. 3, no. 4, pp. 3224–3231, Oct. 2018

work page 2018

[6] [5]

Electric field servoing for robotic ma- nipulation,

R. Wistort and J. R. Smith, “Electric field servoing for robotic ma- nipulation,” in 2008 IEEE/RSJ International Conference on Intelligent Robots and Systems , Sept. 2008, pp. 494–499

work page 2008

[7] [6]

6d proximity servoing for preshaping and haptic exploration using capacitive tactile proximity sensors,

S. Escaida Navarro, M. Schonert, B. Hein, and H. W ¨orn, “6d proximity servoing for preshaping and haptic exploration using capacitive tactile proximity sensors,” in 2014 IEEE /RSJ International Conference on Intelligent Robots and Systems , Sept. 2014, pp. 7–14

work page 2014

[8] [7]

Seashell e ffect pretouch sensing for robotic grasping,

L.-T. Jiang and J. R. Smith, “Seashell e ffect pretouch sensing for robotic grasping,” in 2012 IEEE International Conference on Robotics and Automation , May 2012, pp. 2851–2858

work page 2012