Page 112 - Read Online

Xi et al. Soft Sci 2023;3:26 https://dx.doi.org/10.20517/ss.2023.13 Page 21 of 34

predictions and decisions even in complex and dynamic environments. Zhang et al. designed a TENG,
[203]
which can detect and identify liquid leakage . They designed an intelligent detection and recognition
system based on big data and machine learning technology to identify different liquids, with a classification
accuracy of over 90%. Zhang et al. designed an intelligent mask with a self-powered breathing sensor as the
key component . With the help of machine learning algorithms, the accuracy of distinguishing between
[204]
the healthy group and three groups of chronic respiratory diseases (asthma, bronchitis, and chronic
obstructive pulmonary disease) is as high as 95.5%.

One of the challenges of using machine learning in self-powered IoT sensors is the limited resources
available in terms of processing power and memory. To solve this problem, algorithms can be optimized for
use on low-power devices, and another method is to enhance the output capability of sensors. Another
challenge is the need to train data. In self-powered IoT sensors, data collection is usually limited by available
energy and storage capacity. This requires better training methods and more efficient sensors for assistance.

Application of human-machine interfacing
Electronic skin
An electronic skin, also known as e-skin, is a soft, stretchable, and ultra-thin material that simulates the
properties and functions of human skin . It can detect various stimuli, such as pressure, temperature,
[32]
humidity, and light, and respond to them in a manner similar to human skin [115,127] . The development of
electronic skin is driven by the demand for advanced technologies that can enhance human-computer
interaction and improve medical care . Electronic skins have a wide range of potential applications,
[205]
including artificial limbs, robots, human-computer interfaces, virtual reality, health monitoring, and
[206]
personalized medical treatment . The combination of electronic skins and self-powered sensors opens up
new possibilities for creating intelligent, energy-saving, and self-sustaining equipment and systems. By
integrating self-powered sensors into electronic skins, various physiological and environmental parameters
can be continuously monitored in real time. It can also generate its own energy from environmental energy
(such as body heat, exercise, or sunlight), making the equipment more energy efficient and
sustainable [207,208] . By reducing the need for frequent battery replacement, the self-powered electronic skin
sensor can make the equipment more cost-effective and conducive to environmental protection . Self-
[209]
powered electronic skin sensors provide powerful and multifunctional platforms for new equipment and
[210]
systems . As shown in Figure 8A, Wen et al. developed a crosstalk-free self-powered strain and
[33]
temperature sensing (SPST) sensor . Using the piezoresistive and thermoelectric effects of a conductive
network, SPST sensors can simultaneously detect strain and temperature stimulation and convert them into
independent resistance and voltage signals, respectively. The sensor can be further driven by the thermal
voltage generated under the temperature difference between human skin and the surrounding environment
to realize self-powered temperature and strain sensing. As a wearable electronic device that can be directly
connected to the skin, the SPST sensor can accurately detect the tiny movement of the human body in the
self-powered mode. As shown in Figure 8B, Shi et al. reported a self-powered sensing flexible, breathable,
[206]
and antibacterial TENG electronic skin . It can provide excellent thermal and wet comfort and has a
significant antibacterial effect on Escherichia coli and Staphylococcus aureus. The sensor can sensitively
sense the change of pressure. As shown in Figure 8C, Chen et al. developed a flexible self-powered
temperature and pressure dual-function electronic skin based on triboelectric and thermoelectric coupling
effects . It can effectively convert temperature and pressure stimuli into two independent voltage signals.
[211]
The temperature sensing process is driven by the natural temperature gradient, while the pressure detection
process is driven by the triboelectric effect. It has potential in wearable health monitoring systems and
artificial intelligence perception.