Documente Academic
Documente Profesional
Documente Cultură
to Humans
By RAVINDER DAHIYA
BEST Group, School of Engineering, University of Glasgow, Glasgow G12 8QQ, U.K.
0018-9219 © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Fig. 2. The need to cover 3-D surface of (a) and (b) robot’s body and (c) prosthetic limbs with a large number of touch sensor has been a
key driver application for flexible and conformable e-skin. (d) E-skin is now being used to develop wearable patches (or “second skin”) with
sensors for health monitoring.
A. E-Skin With Off-the-Shelf functional components, such as local been hinting at these advances in
Components memory. However, any new addition 3-D printing [16]. When it comes to
of nonbendable off-the-shelf compo- artificial limbs, new designs will also
To meet the immediate need of nents would severely constrain the be needed to allow routing of wires
tactile feedback in various robotic e-skin in terms of bendability or con- within the 3-D printed structures.
tasks, various e-skin alternatives have formability. In fact, the approach can be extended
been explored, with off-the-shelf sens- The use of FPCB-based approach to any 3-D structure where inherent
ing/electronic components soldered can also be seen in the application, sensing is desired and can open new
on to flexible printed circuit boards such as health monitoring and wear- avenues for electronics packaging and
(FPCBs) [11] or stitched to the flex- able systems, e.g., FPCB-based wear- manufacturing.
ible surfaces [14]. In these cases, able e-skin-like patches to monitor
the e-skin is made of mechani- chronic conditions such as diabetes B. E-Skin With Printed
cally integrated but otherwise dis- [8], [15]. A wide variety of wearable Electronic/Sensing Components
tinct and stiff subcircuit islands of gadgets for wellness applications are Another approach for obtaining
off-the-shelf components connected also available in market today, and e-skin components or patch involves
with wires or stretchable metal inter- the use of FPCB-based sensing patches printing of sensors on flexible
connects. This approach has been can also be seen in fashion. substrates [17], [18]. In fact, printing
explored by various research groups. The additive manufacturing has been widely explored for next-
These semirigid FPCB-based skin explored recently for 3-D PCBs generation electronics. Several tactile
patches conform to surfaces with could offer advances in the FPCB- skin patches with printed (e.g., screen
large curvature, such as arms of based approach. For example, printing and inkjet printing) sensors
humanoid robots (see Fig. 2), and printed sensors embedded in 3-D using composites, such as conductive
have served some of the urgent printed artificial limbs will lead to fillers in the PDMS matrix, have
needs, such as tactile feedback from robust and cost-effective robotic been reported [17]. In these cases,
whole body or large parts of the platforms with inherent tactile the sensitive materials are directly
body. In fact, such large-area imple- sensing (or structurally integrated printed on flexible and soft substrates.
mentations of e-skin have changed sensing), which is also free from The sensors readout on e-skin
the robotics research paradigm—from the traditional wear and tear issue as well as some sensing devices
hand-based manipulation to exploit- associated with e-skin. However, for (e.g., transistors or solid-state sen-
ing multiple contact points or areas this, the 3-D printing techniques must sors) require high-performance elec-
contact to plan and execute robotic evolve from current single material tronics on flexible and conformable
tasks/movements. Further improve- (either metal or polymers) printing to surfaces. The conventional materi-
ments in the FPCB- or stitching-based simultaneous multimaterial printing. als for high-performance electron-
e-skin can be made by including more Some recent works have already ics, such as silicon (Si), are not
of energy harvesters, energy storage may be occurring in humans. For some of the abilities and morphology
devices, low-power electronics, and example, the ensemble of tactile data of real skin with sensors and elec-
efficient/wireless power transfer- from peripheral neurons is consid- tronics embedded in soft substrates.
based technologies, are expected to ered to indicate both the contact force However, just copying skin morphol-
revolutionize the market for wearable and its direction. Therefore, a neu- ogy or capturing few parameters that
systems and, in particular, for e-skin. ronlike inference to handle the tac- we experience as touch is not enough.
In a recent review article [29], we tile data early on can be helpful. The challenge lies in reproducing the
have discussed a wide range of A few works have recently focused functions rather than the shape while
solutions such as lightweight e-skin on such a processing at sensor level accepting the fact that the shape
with wearable energy harvesters (e.g., to detect force and pain [30]–[32]. (morphology), at the micro and macro
photovoltaics, thermoelectric, piezo- Such approaches must be scaled up levels, affects the functions. There-
electric, and triboelectric) and energy for large-area e-skin, even if this is not fore, we must focus on the functional-
storage devices (e.g., flexible batteries an easy task. ities as well, and in this regard, there
and supercapacitors). Among various The software-based neural network is also a need to find the ways to
energy harvesters, the photovoltaics (NN) approaches have also been extract the information from sensor
generate the most and have good explored for tasks, such as object data.
potential for energy-autonomous recognition via texture or materials A meaningful e-skin, for a broad
e-skin—as demonstrated recently [33]. However, the software-based range of applications discussed here,
through graphene-based transparent approach still requires all the data to requires a holistic approach starting
e-skin integrated on photovoltaic be transferred to the central process- from the way the data are acquired,
cells [12]. Wireless powering is ing unit. The NN for inference from encoded, and eventually acted upon.
another attractive alternative, which data gathered by e-skin could work The way ahead lies in multidiscipli-
is possibly more suitable for e-skin for health monitoring of population in nary teamwork: on the one side, neu-
application in health monitoring as a region. The application of software- roscientists and clinicians analyzing
user comfort is important. based approach for e-skin in robotics the multitechnological bases tactile
requires attention as the time between encoding or clinical validity of sen-
data gathering and reflex action sors on e-skin, and on the other side,
E. Neural E-Skin could be very short. The software- engineers and technologists synthesiz-
Efficient ways are needed to based NN approaches are slower ing the artificial systems, not only as
process the sensory data, especially and less energy-efficient due to the “living” proof-of-concept but also scal-
in the case of large-area e-skin where lack of large-scale parallel processing. ing up for real use. Of course, there
a large number of touch sensors are Instead, the hardware-implemented are several challenges in the way.
needed. As an example, a human- neuromorphic tactile data process- Major challenges related to comfort-
inspired e-skin will require about ing along with NNs like algorithms ability, signal acquisition and trans-
45k mechanoreceptors in ∼1.5-m2 could be better alternative. A few mission, and energy autonomy have
area [4], [5]. With whole-body tactile hardware neuromorphic implemen- been discussed in the previous sec-
sensing, the tactile data will increase tations reported in the literature, tions along with the way technology
rapidly and, therefore, will be the although not in context with e-skin, is being developed to overcome some
challenges related to compilation use spin logic [34], memristor, neuron of them. Since various technologies
and processing. Currently, limited MOSFET, analog circuit-based neu- that are being used to develop e-
solutions are available to deal with rons, and field-programmable gate skin are themselves at various stages
large data generated in the tactile array. Although none of these have of development, it may take couple
skin, let alone for the resulting made major impact on robotic e-skin, more years for widespread use of
perception. they could be potential alternative e-skin.
New techniques for the manage- despite challenges in terms of com- Challenges often come with oppor-
ment of the tactile data will add plexity, scalability, speed, reliability, tunities. The opportunity for e-skin
a significant value to the e-skin nonbendability, and power consump- in various application areas has
research. One option is to develop tion. The neural nanowire field-effect been already discussed. In addition,
e-skin with local processing of sensory transistor-based approach is another the development of e-skin is also act-
data, i.e., instead of sensing all raw option for hardware NN [4]. ing as a catalyst for advances in other
data to central unit for processing and areas and is opening up new oppor-
decision-making, sending only the III. C H A L L E N G E S A N D tunities in those areas. For exam-
partially processed data and allow- OPPORTUNITIES ple, with requirement of bendability
ing central unit to take higher level With technological advances over the in e-skin, the conformable electronics
decisions. The neuroscience studies past few years, particularly in the has also advanced and has opened
on human touch sensing suggest that field of flexible and soft electronics, opportunities for developing predic-
such on-site distributed computing we have gotten closer to mimicking tive and accurate models for flexible
devices, which could be exploited to offer unprecedented opportunities for practices at the global scale. E-skin-
predict the bending state or stress pat- tackling several pressing societal chal- type patches are also being explored
terns. This is because device response lenges. For example, when wrapped as smart tags for real-time monitoring
changes constructively or destruc- around surgical tools (e.g., tool used of food and drugs. Coupled with
tively as a result of bending-induced in the pin-hole surgery), the flex- advances in the IoT, e-textiles, virtual
stresses [35]–[37]. ible sensory patches could allow reality, augmented reality, Industry
The e-skin or tactile skin in surgeons the feel tissues or palpate 4.0, organ on chips, and artificial
robotics and “second skin” in internal body parts. The disposable intelligence/deep learning, the e-skin
wearable applications are closely and wearable solutions enabled technologies hold the promise of
linked through flexible electronics by advances in e-skin are aiding achieving a new level of human
technology. Altogether, these areas positive changes in health monitoring connectedness.
REFERENCES
[1] S. Bauer, S. Bauer-Gogonea, I. Graz, skin,” IEEE Sensors J., vol. 1, no. 1, pp. 41–51, pp. 34–40, Jan. 2015.
M. Kaltenbrunner, C. Keplinger, and Jun. 2001. [26] A. Adami, R. S. Dahiya, C. Collini, D. Cattin, and
R. Schwödiauer, “25th anniversary article: [15] W. Gao et al., “Fully integrated wearable sensor L. Lorenzelli, “POSFET touch sensor with CMOS
A soft future: From robots and sensor skin to arrays for multiplexed in situ perspiration analysis,” integrated signal conditioning electronics,” Sens.
energy harvesters,” Adv. Mater., vol. 26, no. 1, Nature, vol. 529, pp. 509–514, Jan. 2016. Actuators A, Phys., vol. 188, pp. 75–81, Dec. 2012.
pp. 149–162, Nov. 2014. [16] H. Ota et al., “Application of 3D printing for smart [27] S. Gupta, D. Shakthivel, L. Lorenzelli, and R.
[2] R. Dahiya, W. T. Navaraj, S. Khan, and E. O. Polat, objects with embedded electronic sensors and Dahiya, “Temperature compensated tactile sensing
“Developing electronic skin with the sense of systems,” Adv. Mater. Technol., vol. 1, p. 1600013, using MOSFET with P (VDF-TrFE)/BaTiO3
touch,” Inf. Display, vol. 31, no. 4, pp. 2–6, 2015. Apr. 2016. capacitor as extended gate,” IEEE Sensors J.,
[3] R. S. Dahiya, G. Metta, M. Valle, and G. Sandini, [17] S. Khan et al., “Technologies for printing sensors vol. 19, no. 2, pp. 435–442, Jan. 2019.
“Tactile sensing-from humans to humanoids,” IEEE and electronics over large flexible substrates: A [28] S. Gupta, W. T. Navaraj, L. Lorenzelli, and R.
Trans. Robot., vol. 26, no. 1, pp. 1–20, Jan. 2010. review,” IEEE Sensors J., vol. 15, no. 6, Dahiya, “Ultra-thin chips for high-performance
[4] W. T. Navaraj et al., “Nanowire FET based neural pp. 3164–3185, Jun. 2015. flexible electronics,” NPJ Flexible Electron., vol. 2,
element for robotic tactile sensing skin,” Frontiers [18] C. G. Núñez et al., Integration Techniques for Mar. 2018, Art. no. 8.
Neurosci., vol. 11, p. 501, Sep. 2017. Micro/Nanostructure-Based Large-Area Electronics [29] C. G. Nunez et al., “Energy autonomous electronic
[5] R. S. Dahiya and M. Valle, Robotic Tactile Sensing: (Elements in Flexible and Large-Area Electronics). skin,” NPJ Flexible Electron., vol. 3, 2019, Art. no. 1,
Technologies and System. Springer, 2012. Cambridge, U.K.: Cambridge Univ. Press, 2018, doi: doi: 10.1038/s41528-018-0045-x.
[6] R. Dahiya, “Electronic skin,” in Proc. 18th AISEM 10.1017/9781108691574. [30] B. C.-K. Tee et al., “A skin-inspired organic digital
Annu. Conf., 2015, pp. 1–4. [19] D.-H. Kim, J.-H. Ahn, W. M. Choi, H.-S. Kim, mechanoreceptor,” Science, vol. 350, no. 6258,
[7] A. Nathan et al., “Flexible electronics: The next T.-H. Kim, J. Song, Y. Y. Huang, Z. Liu, C. Lu, and pp. 313–316, Oct. 2015.
ubiquitous platform,” Proc. IEEE, vol. 100, no. 5, J. A. Rogers, “Stretchable and foldable silicon [31] L. E. Osborn et al., “Prosthesis with neuromorphic
pp. 1486–1517, May 2012. integrated circuits,” Science, vol. 320, no. 5875, multilayered e-dermis perceives touch and pain,”
[8] W. Dang, L. Manjakkal, W. T. Navaraj, L. Lorenzelli, pp. 507–511, Apr. 2008. Sci. Robot., vol. 3, p. eaat3818, Jun. 2018.
V. Vinciguerra, and R. Dahiya, “Stretchable wireless [20] R. Dahiya, G. Gottardi, and N. Laidani, “PDMS [32] C. Bartolozzi, “Neuromorphic circuits impart a
system for sweat pH monitoring,” Biosensors residues-free micro/macrostructures on flexible sense of touch,” Science, vol. 360, pp. 966–967,
Bioelectron., vol. 107, pp. 192–202, Jun. 2018. substrates,” Microelectron. Eng., vol. 136, Jun. 2018.
[9] X. Huang, Bioresorbable Materials and Their pp. 57–62, Mar. 2015. [33] S. Decherchi, P. Gastaldo, R. S. Dahiya, M. Valle,
Application in Electronics (Elements in Flexible and [21] S. Khan, R. S. Dahiya, and L. Lorenzelli, “Flexible and R. Zunino, “Tactile-data classification of
Large-Area Electronics). Cambridge, U.K.: thermoelectric generator based on transfer printed contact materials using computational
Cambridge Univ. Press, 2017, doi: Si microwires,” in Proc. 44th Eur. Solid State Device intelligence,” IEEE Trans. Robot., vol. 27, no. 3,
10.1017/9781108290685. Res. Conf., Sep. 2014, pp. 86–89. pp. 635–639, Jun. 2011.
[10] S.-K. Kang, J. Koo, Y. K. Lee, and J. A. Rogers, [22] K. Takei et al., “Nanowire active-matrix circuitry for [34] J. Grollier, D. Querlioz, and M. D. Stiles, “Spintronic
“Advanced materials and devices for bioresorbable low-voltage macroscale artificial skin,” Nature nanodevices for bioinspired computing,” Proc. IEEE,
electronics,” Accounts Chem. Res., vol. 51, no. 5, Mater., vol. 9, pp. 821–826, Sep. 2010. vol. 104, no. 10, pp. 2024–2039, Oct. 2016.
pp. 988–998, 2018. [23] C. G. Núñez, F. Liu, W. T. Navaraj, A. Christou, [35] H. Heidari, N. Wacker, and R. Dahiya, “Bending
[11] R. S. Dahiya, P. Mittendorfer, M. Valle, G. Cheng, D. Shakthivel, and R. Dahiya, “Heterogeneous induced electrical response variations in ultra-thin
and V. J. Lumelsky, “Directions toward effective integration of contact-printed semiconductor flexible chips and device modeling,” Appl. Phys.
utilization of tactile skin: A review,” IEEE Sensors J., nanowires for high-performance devices on large Rev., vol. 4, no. 3, p. 031101, 2017.
vol. 13, no. 11, pp. 4121–4138, Nov. 2013. areas,” Microsyst. Nanoeng., vol. 4, Aug. 2018, [36] S. Gupta et al., “Device modelling for bendable
[12] C. G. Núñez, W. T. Navaraj, E. O. Polat, and R. Art. no. 22. piezoelectric FET-based touch sensing system,” IEEE
Dahiya, “Energy-autonomous, flexible, and [24] T. Someya, Z. Bao, and G. G. Malliaras, “The rise of Trans. Circuits Syst. I, Reg. Papers, vol. 63, no. 12,
transparent tactile skin,” Adv. Funct. Mater., vol. 27, plastic bioelectronics,” Nature, vol. 540, pp. 2200–2208, Dec. 2016.
p. 1606287, May 2017. pp. 379–385, Dec. 2016. [37] A. Vilouras, H. Heidari, S. Gupta, and R. Dahiya,
[13] D.-H. Kim et al., “Epidermal electronics,” Science, [25] M. Drack, I. Graz, T. Sekitani, T. Someya, “Modeling of CMOS devices and circuits on flexible
vol. 333, no. 6044, pp. 838–843, 2011. M. Kaltenbrunner, and S. Bauer, “An imperceptible ultrathin chips,” IEEE Trans. Electron Devices,
[14] V. J. Lumelsky, M. S. Shur, and S. Wagner, “Sensitive plastic electronic wrap,” Adv. Mater., vol. 27, vol. 64, no. 5, pp. 2038–2046, May 2017.