Professional Documents
Culture Documents
June 2023
WRITTEN BY:
Rotem Sokolovsky
305706368
This work was carried out under the supervision of:
• The fingers were designed with a structure that can withstand the forces and pressures
exerted during manipulations.
• The fingers were simple to prepare in laboratory conditions and were low-cost, making
them accessible to a wider range of researchers and developers.
• The round structure of the fingers allowed for complex manipulations to be performed
inside the robotic hand.
1. Table of Contents:
1 Abstract .....................................................................................................................2
1.1 Main System Requirements: ............................................................................2
1.2 Project Overview:.............................................................................................2
1.3 Project Achievements- Measuring Performance Against Requirements...........2
1. Table of Contents:.....................................................................................................3
2. List of Figures ...........................................................................................................4
3 Introduction ...............................................................................................................5
4 Project Requirements ...............................................................................................6
5 Related Work .............................................................................................................6
5.1 Pressure Sensor ..............................................................................................6
5.2 Vision-Based Tactile Sensing ..........................................................................7
5.2.1 Optical Tactile Sensors.........................................................................7
5.2.2 GelSight Sensors .................................................................................7
6 Design ........................................................................................................................8
6.1 Motivation ........................................................................................................8
6.2 Previous version of the robotic finger ...............................................................9
6.2.1 Silicon Shell ..........................................................................................9
6.2.2 Rigid finger .........................................................................................10
6.2.3 LEDs PCB ..........................................................................................10
6.2.4 Main Body ..........................................................................................10
6.2.5 Bottom cap .........................................................................................11
6.3 Research and Development Process .............................................................11
6.3.1 Rigid Finger ........................................................................................11
6.3.2 Silicon Shell ........................................................................................16
6.3.3 Camera Holder ...................................................................................20
7 Preparation Method ................................................................................................21
7.1 Equipment List / Simulation Tools ..................................................................21
8 Procedure ................................................................................................................23
8.1 3-D prints - Use SLA printer ...........................................................................23
8.2 Post-Printing .....................................................Error! Bookmark not defined.
8.3 Electrical Design ............................................................................................25
8.4 Software- Camera Streaming.........................................................................25
8.5 Assembly .......................................................................................................26
9 Result and discussion ............................................................................................27
10 Bibliography ............................................................................................................28
2. List of Figures
Figure 1: The robotic finger touches a screw (left), The resulting image on the screen
(right)............................................................................................................................ 5
Figure 2: Previous version of the robotic finger from REF 30. .............................................. 9
Figure 3: Silicon Shell (REF 30)............................................................................................... 9
Figure 4: Rigid finger ) REF 30)............................................................................................. 10
Figure 5: Printed PCBA (left), Printed PCB (right), (REF 30) ................................................ 10
Figure 6: Main body connections (REF 30) .......................................................................... 11
Figure 7: Main body connection base (REF 30) ................................................................... 11
Figure 8: Rigid finger made of epoxy (left), complex shape SLA printed shell (Right) ........ 12
Figure 9: FDM part on the left, SLA part on the right REF 35). ........................................... 13
Figure 10: Flashforge Hunter SLA printer (left), Creality UW-02 curing machine (right).... 14
Figure 11: The printing material remains trapped inside the Shell. ................................... 14
Figure 12: Supports inside the shell after the print (left), an illustration of the shell before
printed FlashDLprint printing software (right)........................................................... 15
Figure 13: Various shell drills are shown in the FlashDLprint printing software ................ 15
Figure 14: upside down print position of the Rigid finger .................................................. 15
Figure 15: Silicon shell before painting and the camera view when touching the shell (REF
30) .............................................................................................................................. 17
Figure 16: dipping in silicone paint ...................................................................................... 18
Figure 17: Silicon shell after dipping in paint mixture with ecoflex material ..................... 18
Figure 18: the markers make diverse motion field patterns under different kinds of forces
or torques. The magnitude of the markers’ motion is roughly proportional to the
force/torque value. REF 11) ....................................................................................... 19
Figure 19: Mr. Osher Azulay demonstrates a change in the silicon shell when touching it
with different objects................................................................................................. 19
Figure 20: gripper connector (left), bottom cap (middle), main body (right) .................... 20
Figure 21: gripper connector revision 02 (left), Collimator (right) ..................................... 20
3 Introduction
The robotics industry has been rapidly evolving in recent years, leading to increased use of robots
across various industries. However, one of the key challenges in this field is developing robots that
can perform complex palm manipulations with objects with the same level of precision and skill as
humans. This requires the development of complex robotic grippers composed of fingers with
sensing capabilities to detect pressures, deformations, and shapes.
There are various types of touch sensors, including resistance, capacitance, optics, and
piezoelectricity, each with its unique characteristics and applications. However, for reliable and easy-
to-use sensors with high resolution and sensitivity, vision-based sensors have become increasingly
popular. Although some fingers with these types of sensors are available on the market, only a few
can perform complex manipulations within the hand.
To achieve these manipulations in an environment where the robot can determine how, when, and
where it interacts with different objects, accurate measurement of the forces and pressures applied
during contact between the robotic finger and the object is also a challenge.
In this report, the application of Vision-Based Tactile sensing technology is explored. This technology
enables a representative of a robotic finger to convert deformations on an external dome into force.
Figure 1: The robotic finger touches a screw (left), The resulting image on the screen (right)
4 Project Requirements
The project aims to develop a prototype that integrates electrical and structural elements for high-
quality image processing in hand manipulations using adaptive grippers. To achieve this, the
following project requirements have been established:
• Geometry - The finger should have a domed structure that allows for 360-degree image
analysis. It should have a round cross-section with a maximum diameter of 40 mm and a
maximum height of 60 mm (without a connection interface to the clamp).
• Stable and durable structure - The durability of the product must be tested with an
experiment that focuses on resistance to multiple pressures on the finger with varying
intensities.
• Modularity - The finger structure should be designed for maximum modularity, with
minimal dependence on external factors that come in fixed sizes. It should be easy to
adjust the finger's dimensions to accommodate different types of grips.
• Integration with gripper - The finger should have a convenient interface for connection to
a robotic gripper that allows for easy access to connect and disconnect the electronics.
The connection to the gripper should be modular to accommodate different types and
sizes of clamps.
• Cost - The project should focus on reducing the costs involved in the production of a
robotic finger with a camera built into the body of the finger. The manufacturing
processes should be minimized to reduce production costs. Cheap electronic products
should be used as much as possible while maintaining quality. The planning of printed
circuits should be done internally without the participation of external parties except for
production and order.
• Assembly instructions - Instructions for the preparation and assembly of a finger should
be prepared for the home consumer after the design phase is completed and a prototype
is developed.
5 Related Work
This chapter provides an extensive review of the existing literature on tactile sensing technologies,
focusing specifically on pressure sensors and vision-based tactile sensing. These two sensor types
have gained significant attention in recent years and are crucial areas of research for developing
robots with advanced tactile sensing capabilities like those of humans.
Initially developed for applications in manufacturing and research and development processes,
pressure sensor arrays have been adapted for use in robots, particularly in the development
of soft sensors that emulate the signals generated by mechanoreceptors in human fingertip
skin. One notable example is the SynTouch BioTac sensor [REF 1]. This sensor comprises an
elastic skin filled with an incompressible conductive fluid, thermistors, impedance sensing
electrodes, and hydro-acoustic pressure sensors. It can detect normal and tangential forces, as
well as the thermal conductivity of objects in contact. The BioTac sensor, known as proximity,
contact, and force (PCF) sensor, incorporates an infrared proximity sensor embedded in a
transparent elastomer to estimate object proximity and applied force at the fingertip.
Various other types of tactile sensors based on different transduction mechanisms exist,
including strain gauges, resistive/piezoresistive sensors, ultrasound sensors, tunnel effect
sensors, impedance sensors, piezoelectric sensors, and magnetic sensors [REF 2].
Vision-Based tactile sensors convert signals of the contact deformation into images, achieving
high spatial resolution and high sensitivity to the contact force. These sensors capture the
deformation of the contacting surface using a camera and infer important information such as
the shape of the detected object, the shear force, and the torque
Texture-based methods incorporating arrays of dots or grids on the contact surface have also
been implemented [REF 6, REF 7, REF 8, REF 9] to encode edge information, and reconstruct
surface traction fields and force vector fields. However, these methods have limitations in
reconstructing high-resolution 3D maps due to the large and sparse texture on the surface,
which restricts the achieved spatial resolution compared to imaging systems
Several designs have been developed over the years: Gelsight by Dong et al. [REF 11], DIGIT by
Lambeta et al. [REF 12] which introduced a low-cost and compact tactile sensor designed for
in-hand manipulation, Gelslim Donlon et al. [REF 13], and GelSlim 3.0 by Taylor et al. [REF 14]
which offers force, slip, and depth estimation. Many of these designs have focused on
miniaturization and have been extensively explored for a wide range of applications by
researchers such as Cao et al. [REF 15, REF 16 ], Jiang et al. [REF 17], Lee et al. [REF 18], Jing et
al. [REF 19], Zhao et al. [REF 20], Calandra et al. [REF 21].
However, most of these designs are limited to a single flat sensing area, which restricts their
potential for use in unstructured environments where unexpected contacts can occur both
within and outside grasp closures. To overcome this limitation, recent designs have proposed
the use of highly curved and/or domed finger-shaped surface membranes. Examples include
the GelTip sensor proposed by Gomes et al. [REF 23, REF 22], the OmniTact sensor developed
by Padmanabha et al. [REF 24], and a semi-round sensor introduced by Romero et el. [REF 1]
While GelTip [REF 23, REF 22] and Romero et al. [REF 1] follow the original GelSight sensor
design by employing glass specters to guide light through the membrane, the OmniTact sensor
[REF 24] directly illuminates the membrane surface, a technique also applied in flat sensors
such as those by Donlon et al. [REF 13] and Lambeta et al. [REF 12]. This approach ensures a
more homogeneous light distribution and enables the use of methods like Poisson surface
reconstruction for accurate elastomer surface reconstruction.
Some of the uses of the GelSight sensors are shown by Li et al. [REF 25] who used the GelSight
sensor to recognize 40 different surface textures. Jia et al. [REF 26] demonstrated that the
GelSight sensor outperformed humans in detecting lumps in soft media, indicating its potential
application in diagnosing breast cancer. Li et al. [REF 27] designed a fingertip GelSight tactile
sensor that significantly reduced the sensor's volume and equipped it on a Baxter robot hand,
which completed a USB insertion task. Yuan et al. [REF 28] further improved the sensor by
adding markers on the gel surface. By analyzing the marker motion, the GelSight sensor can
sense normal, shear, and torsional loads on the contact surface, as well as detect incipient slip.
The GelSight sensor has also been used to detect the hardness of contact objects based on the
analysis of gel deformation and marker displacement in [REF 29].
6 Design
6.1 Motivation
The motivation behind the new design is to address the shortcomings of the previous
prototype of a robotic finger with optical sensing capabilities. The initial design did not meet
the durability and transparency requirements, and the new design aims to improve upon these
aspects. The main system requirements for the new design include being small, durable, low-
cost, easy to assemble and operate, and having a transparent and uniform photo space for
high-quality image capture and analysis.
Figure 2: Previous version of the robotic finger from Error! Reference source not found.
The silicon shell on the previous revision of the robot finger coloring did not sufficiently block
light, resulting in inaccurate and low-resolution force analysis.
In summary, the previous design of the robotic finger had several shortcomings, including weak
connections, complicated production processes, unsatisfactory transparency, and inaccurate
force analysis. The new design aims to address these issues and meet the project's
requirements.
o High-pressure resistance: eSUN transparent resin, which is specifically designed for SLA
printing, has high tensile strength, 46-67 [MPa], and a modulus of elasticity of 1000-
1400 [MPa]. In comparison, the tensile strength of epoxy resins can range from 45
[MPa].
Table 1: eSun resin's (SLA print) parameters compare to EpoxAcast 690's parameters
o High accuracy: SLA technology provides the tightest tolerances of any 3D printing
technology on the market. The heated resin tank and contained build environment
ensure near-identical conditions for each print. The process also occurs near room
temperature, meaning SLA parts and prototypes aren’t as affected by thermal
contraction and expansion compared to other manufacturing methods, leading to a
high level of accuracy and repeatability. When using the SLA printer to create the shell,
the print precision is ±0.05mm.
Table 2: SLA printer accuracy parameters compares to FDM printer accuracy parameters
Flashforge FOTO 8.9s SLA 3D Printer (REF 34) Flashforge Guider II FDM 3D Printer (REF 33)
o Improved surface quality: SLA 3D printing is known for its exceptional ability to produce
high-quality, smooth surface finishes, making it the top choice for additive
manufacturing. The surface finishes generated by SLA 3D printing are comparable to
those achieved by injection molding, CNC machining, and extrusion techniques. The
smooth surface finishes are particularly advantageous in reducing post-processing time,
as the prints come out of the printer with a smooth finish. Unlike other 3D printing
methods, the interaction between the most recent layer printed and the previous layer
in SLA printing smooths out the typical staircase effect, resulting in a refined surface
finish
When creating a flexible pattern, it adopts the surface quality of the object that contains
it. If the object was produced using FDM printing, the resulting surface quality of the
pattern will be in layers. Consequently, the Rigid finge also inherits this layered surface
quality. In contrast, SLA printing produces a smoother surface quality compared to FDM
printing when printing the shell. This means that an SLA-printed Rigid finger will have a
superior surface quality compared to an FDM-printed one.
Figure 9: FDM part on the left, SLA part on the right REF 35).
o Easy to create: Creating objects with an SLA printer is a straightforward process that is
easy to learn. The printing process itself takes around 3 hours, which is much faster
than creating objects with epoxy. Once the printing is complete, the shell requires only
sanding, a process that can be completed immediately. Furthermore, since the printing
time is determined by the height of the object rather than the number of objects,
many shells can be printed without increasing the printing time.
In contrast, creating objects with epoxy casting requires printing multiple molds to
contain silicone castings, which then create another mold for the epoxy casting.
Additionally, working with epoxy requires a significant amount of knowledge and
precise adherence to the creation procedure, which can compromise the modularity of
the system.
• Print direction
After agreeing that printing the parts with SLA technology would improve the design, it
was necessary to define the printing characteristics.
Platform
Curing
accelerator
Vat / container
Figure 10: Flashforge Hunter SLA printer (left), Creality UW-02 curing machine (right)
Due to the form of printing, material may be trapped inside a closed body when the
part is cured on top of the sides of the part something that damages the symmetry of
the part and the uniform passage of light, which happened during the printing of the
Rigid finger, as you can see in Figure 11 Figure 10.
Platform
Figure 11: The printing material remains trapped inside the Shell.
In addition, this printing process requires many supports for the part to remain fixed to
the platform that rises and falls, as a result, a situation is created in which there are
many supports inside the shell, which also affects the uniformity and symmetry of the
part and the passage of light out.
Platform
Figure 12: Supports inside the shell after the print (left), an illustration of the shell before printed
FlashDLprint printing software (right).
Figure 13: Various shell drills are shown in the FlashDLprint printing software
• Sanding the inside of the shell with sandpaper, led to an unsatisfactory result.
• Printing the part upside down: although supports were created on the outer side
of the shell, access to them during the sanding of the part was easy and material
did not accumulate inside the shell
Sanding method-
sandpaper
Sanding method-
sponge
• Lacquer
To prevent soiling of the Rigid finger, it was decided to add lacquer on top of the part. This
action and repeating it several times led to the required transparency. First, the process
was carried out with matte clear lacquer which, although it made the part transparent,
gave it a yellow tint. After that, it was decided to replace it with a glossy clear lacquer
which gave the required transparency.
Table 4: result for the printed silicon shell by using a casting mold that printed in 2 different technologies
Print Silicon Shell Casting Silicon Shell Casting Result before Result after
technology Mold-Bottom Mold-Top painting painting
FDM NA
SLA
Figure 15: Silicon shell before painting and the camera view when touching the shell
(REF 30)
Therefore, it is necessary to produce a layer of paint for blocking External light for the
finger. In the previous design, the coating is made of a silicone paint base, Smooth-on
Psycho Paint. The painting was done by dipping it in silicone paint and waiting for drying.
Painting in this way caused a lot of waste of material due to its dripping and in addition,
the result did not adequately seal the light that is outside the silicon shell.
Figure 16: dipping in silicone paint
In the revision 2 design, the color was changed from white to gray and after drying the
wound, another immersion of the silicon shell was carried out in a mixture to which
ecoflex material was added to strengthen the adhesion to the silicon.
Figure 17: Silicon shell after dipping in paint mixture with ecoflex material
In the following table, you can see the stages of light blocking by the color after each
immersion
Figure 19: Mr. Osher Azulay demonstrates a change in the silicon shell when touching it with different
objects
Table 5: Blocking the lighting in different stages of painting
o Main Body, which held the Rigid finger, the silicon shell, and the PCB leds. In addition, its
function was to allow optimal resolution by maintaining a distance of X from the camera.
o Bottom Cap which is designed to hold the camera in place
o Gripper Connector connects all the parts to the robotic arm.
Figure 20: gripper connector (left), bottom cap (middle), main body (right)
In revision 02, these parts were combined into one part, which was made possible thanks to
the switch to SLA printing. In addition, after examining revision 01 under large loads of the
system, it was decided that there is a need to connect the parts of the system rigidly with
screws instead of connecting them with snaps. Moreover, it was decided to add a collimator
to route the LED lighting uniformly through the finger. Since in the examination of revision01,
it was noticed that the dedicated glue that sticks to the camera did not hold the camera
sufficiently, it was decided to add a notch to the camera mount to fix the camera in place.
Raspberry Pi
2. Raspberry Pi NA
Zero1.3/W or 4B
High Transparent
5. Acrylic Resin
(Datasheet REF 36)
Or similar
7. Solaris Kit
Silicon rubber
Ecoflex
8.
(Datasheet, REF 41)
Smooth-On
Plastic Metallic
Metallic Spray Krylon Shimmer Aerosol
13. NA
Paint Spray Paint.
or similar
8 Procedure
Material
Part Infill [%] Printing Position Pictures
(From Section 7.1)
Camera
4 100
holder
Collimator 4 100
1.
Casting mold
4 40-50
(bottom+top)
8.2 Post-Printing
Step Description Notes
2. Collimator
3. Rigid finger
Grind the outer surface of the finger to get rid of the
4. Gel
Spray 2 layers of lacquer (Section 7.1, 13) on the inner
4.1 side of the bottom mold and the top mold. Wait at least NA
2 hours after each layer.
Make the gel mixture: 6 gr of Solaris (Section 7.1, 13) for
Put the bottom mold with the gel in a pressure box (of
Put the top mold in the gel and make sure it is placed
well. Place a high-weight plate on the mold to prevent
4.5
movements of the top part. Wait at least 2 days before
removing it.
RRRGGGBBB
RGBRGBRGB
White
Brand Link
Raspberry Pi Zero1.3 link
8.5 Assembly
Step Description Notes
6. Assembly
smear epoxy glue (Section 7.1, 9) on the outer side of
6.1 the rigid finger. Gently push it into the gel until there are
no air gaps. Let it dry for at least 1 day.
6.2 Put 3 M1.6 heat inserts in the holes of the camera holder
In conclusion, the proposed design of a cheap, compact, and high-resolution tactile sensor for
manipulations inside a robotic hand has been successful in meeting the project's requirements. The
improved design is cost-effective and easily assembled, making it accessible to researchers and
developers. Future work can involve testing the finger's integrity, durability, and image processing
capabilities and integrating it into a complete robotic hand for practical applications.
10 Bibliography
REF 1 Romero, B. R. (2017). Soft, round, high-resolution tactile fingertip sensors for dexterous robotic
manipulation. University of Colorado Boulder.
REF 2 Dahiya, R. S., Metta, G., Valle, M., & Sandini, G. (2009). Tactile sensing—from humans to
humanoids. IEEE Transactions on Robotics, 26(1), 1-20.
REF 3 Jiar, Y., Lee, K., & Shi, G. (1993). A high-resolution and high-compliance tactile sensing system for
robotic manipulations. In Intelligent Robots and Systems '93, IROS '93. Proceedings of the 1993
IEEE/RSJ International Conference on (Vol. 2, pp. 1005–1009). IEEE.
REF 4 Ohka, M., Mitsuya, Y., Hattori, K., & Higashioka, I. (1996). Data conversion capability of an
optical tactile sensor featuring an array of pyramidal projections. In Multi-sensor Fusion and
Integration for Intelligent Systems, 1996. IEEE International Conference On (pp. 573–580). IEEE.
REF 5 Ferrier, N. J., & Brockett, R. W. (2000). Reconstructing the shape of a deformable membrane
from image data. The International Journal of Robotics Research, 19(9), 795–816.
REF 6 Chorley, C., Melhuish, C., Pipe, T., & Rossiter, J. (2009). Development of a tactile sensor based on
biologically inspired edge encoding. In Advanced Robotics, 2nd International Conference on (pp.
1–6). IEEE.
REF 7 Sato, K., Kamiyama, K., Nii, H., Kawakami, N., & Tachi, S. (2008). Measurement of force vector
field of a robotic finger using vision-based haptic sensor. In Intelligent Robots and Systems,
2008. IRO 2008. IEEE/RSJ International Conference on (pp. 488–493). IEEE.
REF 8 Nagata, K., Ooki, M., & Kakikur, M. (1999). Feature detection with an image-based compliant
tactile sensor. In Intelligent Robots and Systems, 1999. IROS'99. Proceedings. 1999 IEEE
International Conference (Vol. 2, pp. 838–843). IEEE.
REF 9 Kamiyama, K., Vlack, K., Mizota, T., Kajimoto, H., Kawakami, K., & Tachi, S. (2005). Vision-based
sensor for real-time measuring of surface traction fields. IEEE International Conference
Applications, 25(1), 68–75.
REF 10 Johnson, M. K., & Adelson, E. H. (2009). Retrographic sensing for the measurement of surface
texture and shape. In IEEE CVPR (pp. 1–8).
REF 11 Yuan, W., Dong, S., & Adelson, E. (2017). Gelsight: High-resolution robot tactile sensors for
estimating geometry and force. Sensors, 17(12), 2762.
REF 12 Lambeta, M., Chou, P.-W., Tian, S., Yang, B., Maloon, B., Most, V. R., ... Calandra, R. (2020).
DIGIT: A novel design for a low-cost compact high-resolution tactile sensor with application to
in-hand manipulation. IEEE Robotics and Automation Letters (RA-L), 5(3), 3838–3845.
REF 13 Donlon, E., Dong, S., Liu, M., Li, J., Adelson, E., & Rodriguez, A. (2018). Gelslim: A high-resolution,
compact, robust, and calibrated tactile-sensing finger. In 2018 IEEE/RSJ International Conference
on Intelligent Robots and Systems (IROS) (pp. 1927–1934). IEEE.
REF 14 Taylor, I. H., Dong, S., & Rodriguez, A. (2022). Gel-Slim 3.0: High-resolution measurement of
shape, force, and slip in a compact tactile-sensing finger. In IEEE ICRA (pp. 10781–10787).
REF 15 Cao, G., Zhou, Y., Bollegala, D., & Luo, S. (2020). Spatio-temporal Attention Model for Tactile
Texture Recognition. In IEEE IROS (pp. 9896–9902).
REF 16 Cao, G., Jiang, J., Mao, N., Bollegala, D., Li, M., & Luo, S. (2023). Vis2hap: Vision-based haptic
rendering by cross-modal generation. In ICRA.
REF 17 Jiang, J., Cao, G., Gomes, D. F., & Luo, S. (2021). Vision-guided active tactile perception for crack
detection and reconstruction. In 29th Mediterranean Conference on Control and Automation
(MED) (pp. 930–936).
REF 18 Lee, J.-T., Bollegala, D., & Luo, S. (2019). "Touching to see" and "seeing to feel": Robotic cross-
modal sensory data generation for visual-tactile perception. In ICRA (pp. 4276–4282). IEEE.
REF 19 Jing, X., Qian, K., Jianu, T., & Luo, S. (2023). Unsupervised adversarial domain adaptation for sim-
to-real transfer of tactile images. IEEE Transactions on Instrumentation and Measurement.
REF 20 Zhao, Y., Jing, X., Qian, K., Gomes, D. F., & Luo, S. (2023). Skill generalization of tubular object
manipulation with tactile sensing and sim2real learning. Robotics and Autonomous Systems,
160, 104321.
REF 21 Calandra, R., Owens, A., Jayaraman, D., Lin, J., Yuan, W., Malik, J., Adelson, E. H., & Levine, S.
(2018). More than a feeling: Learning to grasp and regrasp using vision and touch. IEEE RA-L,
3(4), 3300–3307.
REF 22 Gomes, D. F., Lin, Z., & Luo, S. (2020). Blocks world of touch: Exploiting the advantages of all-
around finger sensing in robot grasping. Frontiers in Robotics and AI, 7, 127.
REF 23 Gomes, D. F., Lin, Z., & Luo, S. (2020). Gel-tip: A finger-shaped optical tactile sensor for robotic
manipulation. In IROS/RSJ.
REF 24 Padmanabha, A., Ebert, F., Tian, S., Calandra, R., Finn, C., & Levine, S. (2020). Omnitact: A multi-
directional high-resolution touch sensor. In IEEE ICRA (pp. 618–624).
REF 25 Li, R., & Adelson, E. (2013). Sensing and recognizing surface textures using a Gelsight sensor. In
Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 1241–
1247).
REF 26 Jia, X., Li, R., Srinivasan, M. A., & Adelson, E. H. (2013). Lump detection with a Gelsight sensor. In
World Haptics Conference (WHC), 2013 IEEE (pp. 175–179).
REF 27 Li, R., Platt, R., Yuan, W., ten Pas, A., Roscup, N., Srinivasan, M. A., & Adelson, E. (2014).
Localization and manipulation of small parts using Gelsight tactile sensing. In Intelligent Robots
and Systems (IROS 2014), 2014 IEEE/RSJ International Conference on (pp. 3988–3993)
REF 28 Yuan, W., Li, R., Srinivasan, M. A., & Adelson, E. H. (2015). Measurement of shear and slip with a
Gelsight tactile sensor. In Robotics and Automation (ICRA), 2015 IEEE International Conference
on (pp. 304–311).
REF 29 Yuan, W., Srinivasan, M. A., & Adelson, E. H. (2017). Gelsight: High-resolution robot tactile
sensors for estimating geometry and force. Sensors, 17(12), 2762.
REF 30 Shlomovitz, D., Liling, G. H., Azulay, O. (2022). אצבע רובוטית עם יכולת חישה אופטית למניפולציות
רובוטיות. Tel Aviv University
REF 31 Smooth-On. EpoxAcast 690/692 Technical Bulletin https://www.smooth-
on.com/tb/files/EPOXACAST_690_692_TB.pdf
REF 32 Spider3D. Silicone Resin for DLP/SLA Printers
REF 33 FlashForge Guider 2 3D Printer https://www.flashforge.com/product-detail/flashforge-guider-2-
3d-printer
REF 34 FlashForge Foto 89S LCD 3D Printer https://www.flashforge.com/product-detail/flashforge-foto-
89s-lcd-3d-printer
REF 35 Formlabs. FDM vs SLA: Compare the Types of 3D Printers https://formlabs.com/asia/blog/fdm-
vs-sla-compare-types-of-3d-printers/
REF 36 Alma Standard Rigid Resin Data Sheet- https://3dny.co.il/wp-
content/uploads/2022/03/%D7%9E%D7%A4%D7%A8%D7%98-
%D7%AA%D7%A4%D7%A2%D7%95%D7%9C-
%D7%97%D7%95%D7%9E%D7%A8%D7%99%D7%9D-ALMA-RESIN.pdf
REF 37 Smooth-On. (n.d.). Ecoflex Series Technical Bulletin https://www.smooth-
on.com/tb/files/ECOFLEX_SERIES_TB.pdf
REF 38 Smooth-On. NOVOCS Technical Bulletin https://www.smooth-on.com/tb/files/NOVOCS_TB.pdf
REF 39 Smooth-On. (n.d.). Sil-Poxy Silicone Adhesive Technical Bulletin https://www.smooth-
on.com/tb/files/Sil_Poxy_Silicone_Adhesive.pdf
REF 40 Raw Material Suppliers. (n.d.). Gray Silicone Ink
https://www.rawmaterialsuppliers.com/product/gray-silicone-ink/
REF 41 Show-me Webcam. (n.d.)., https://github.com/showmewebcam/showmewebcam