Development of electrically poled NBT single crystal based FSR sensor for smart posture monitoring applications

1 Introduction

Recently, the rapid progress of technology and changing environmental conditions have considerably accelerated the acceptance of work-from-home and computer-based professional activities. Employees who work in the information technology industries, financial sectors, software firms, and other offices spend long hours in front of computer systems with inappropriate or unchangeable sitting postures. Prolonged sedentary work and the wrong posture in a chair can cause damage to the human body, notably the spinal cord and the muscles of the neck, shoulders and lower back. These problems can develop gradually and lead to musculoskeletal illnesses, distortion of the backbone, weariness, stress, and low physical wellbeing (Dunne et al., 2008; Yang et al., 2008; Walsh et al., 2006; Jang et al., 2020).

The poor ergonomic conditions and limited physical movement that many office and remote employees experience are causing a rapid rise in health diseases connected to posture, as cited by WHO report, 619 million people globally were affected by low back pain in 2020, further, it is predicted that it will increase to 843 million by 2050 (Ferreira et al., 2023). Thus, intelligent healthcare-support systems that continually monitor sitting positions have become an important study subject in embedded systems and sensor technologies. Smart posture monitoring systems based on embedded electronics and pressure-sensing technologies have attracted much research interest as a solution to these problems. Force-sensitive resistor (FSR) sensors have been widely employed in the healthcare monitoring field because the piezoelectric response and mechanical strength show that the produced NBT single crystal has promising features for fabrication, such as its small size, high sensitivity, low power consumption, and real-time response qualities. The mechanical stability and electromechanical conversion efficiency of the sensing material primarily govern the performance of FSR sensors (Wang et al., 2020; Ferreira et al., 2023; Li et al., 2026; Kim et al., 2018; Zhang et al., 2021; Li et al., 2024; Zemp et al., 2016; Gyi and Porter, 1999).

In the present work, a slow bottom-cooling flux growth process harvested NBT single crystals, and they have demonstrated excellent potential in FSR sensor fabrications. The generated single crystals were electrically poled for better ferroelectric domain alignment and piezoelectric responsiveness. The Vickers microhardness measurements were conducted to assess the relationship between piezoelectric response and mechanical strength, indicating that the produced NBT single crystal has promising characteristics for fabricating pressure and force sensing applications. The piezoelectric coefficient d₃₃ was acquired and results verified enhanced electromechanical charge production capability under applied force, while the hardness study indicated enhanced mechanical stability following electrical poling. The association between piezoelectric response and mechanical strength indicates that the produced NBT single crystal has promising features for fabricating extremely sensitive and mechanically reliable FSR sensors. The association between piezoelectric response and mechanical strength indicates that the produced NBT single crystal has promising features for fabricating extremely sensitive and mechanically reliable FSR. sensors. The developed embedded posture monitoring system is manufactured using FSR sensors to detect poor sitting posture and provides corrective feedback to the user in real time. The method created can be useful for the prevention of backbone-related ailments, improving ergonomic setup in the workplace and for preventive healthcare applications in offices, homes, hospitals and rehabilitation facilities. Furthermore, the integration of the embedded controller, real-time sensing mechanisms, and intelligent monitoring approaches into the proposed system enhances its robustness, making it a stronger option for next-generation smart healthcare and ergonomic aid applications (Cucchiara et al., 2005; Juang et al., 2007; Zheng et al., 2010; Jiang et al., 2022; Sifuentes et al., 2019; Banga et al., 2023; Diraco et al., 2014; Diraco et al., 2013; Ramesh Babu et al., 2017; Hussain et al., 2022; Rodríguez de Trujillo et al., 2018; Gaiduk et al., 2020; Huang et al., 2017).

2 Experimental procedure

Na₂CO₃, Bi₂O₃ and TiO₂ of high purity (≥99.9%) were utilised as starting chemicals. In the first stage, the single-phase NBT powder was synthesised by the traditional solid-state approach. Then the synthesised powder was combined with flux, i.e., Bi₂O₃ and Na₂CO₃, in the 1:1 stoichiometric ratio. The mixed reactants were enclosed in a platinum crucible and were immersed in fine TiO₂ powder in an alumina crucible, to prevent undesirable evaporation of bismuth and sodium oxides at high temperature. Figure 1 shows that material synthesis and crystal growth flow path (Kiriazi et al., 2021; Xu et al., 2023; Xiang et al., 2024).

Crystal formation was completed out in an automated furnace with resistive heating and temperature controls that used a bottom cooling method. The crystals were grown in a resistive heating box furnace at 1250 °C for 5 h, then cooled to 1200 °C at a rate of 5 °C per hour, and finally chilled at 50 °C per hour. Several crystals were discovered to form in the crucible’s center and along the wall. The formed crystals were washed out of the platinum crucible with 20% HNO₃ acid above room temperature through boiling from a residual flux. Some of the crystals were broken into powder and tested for phase identification. Figure 2 depicts how the single crystal was harvested and sliced into six rectangular pieces (10 mm × 15 mm) using a wire cutter, and then all of the crystal pieces were polished and transformed into electrodes by pasting them with a silver electrode, resulting in sensors. An electric field of 3 kV/mm was applied to all NBT single crystals at 100 °C for 30 min following electrode deposition to electrically pole them. Adjusting the molar concentration of the KNN solution was the first step in achieving the ideal poling conditions and the required d₃₃ response. The analytical results showed that for the proposed FSR sensor application, a concentration of 5 mol% KNN yields the best d₃₃ response. The crystals were cooled to room temperature once polishing was finished and the applied electric field was removed (Claus et al., 2016; Cho et al., 2015; Grandner, 2017).

The proposed system’s structure has been designed to identify eight different sitting postures, using a sensor-equipped office chair. The overall design of the proposed system is shown in Figure 3. The office chair’s hardware module comprises a FSRs, ToF sensors, and an ESP32 microcontroller. The collected data is processed on a laptop, and the sitting posture is identified by a rule-based classifier.

The block diagram (Figure 4) depicts the interdigitated silver electrode based flexible pressure sensor pad, VL53L0X, and ESP32 as the system’s FSR, ToF, and microcontroller, respectively. The FSR detection range is 0.2 N to 20 N, while the actuation force is 0.2 N. The VL53L0X is the ToF sensor, capable of measuring distances of up to 8 m. The ESP32 microcontroller is used to control and monitor the system. The sensors were integrated with an ESP32 and connected to a laptop (Chattu et al., 2019; Rundo and Downey, 2019).

The hardware circuit (Figure 5) is constructed with 6 FSRs and 2 ToF sensors connected to an ESP32 controller. The controller’s 3.3 V and GND connectors deliver power to the ToF sensors. To facilitate I2C connectivity, the ToF sensor’s SCL and SDA pins are linked to the controller’s SCL and SDA pins. The first terminal of each FSR sensor was connected to the 3.3 V power pin, while the second terminal was connected in parallel to a 10 kΩ resistor and the ADC pin. Connecting the opposite end of the 10 kΩ resistor to the ground pin created a voltage divider circuit. The ESP32 is then powered by either the laptop’s USB port or the battery cell. Figure 6 shows the interconnection of components of the proposed system.

3 Results and discussion

3.1 Structural characterization

Figure 7 shows powder x-ray diffraction pattern of the formed single crystals powder. X-ray diffraction reveals single-phase formation with no additional phase, according to Joint Committee on Powder Diffraction Standards (JCPDS) file number 25–1126. The diffraction peaks were indexed based on the cubic perovskite structure identified in the ferroelectric phase. In addition to the fundamental perovskite reflections, a super lattice reflection near 38° was identified. The presence of super-lattice reflection is caused by the existence of a rhombohedral structure with the R3c (161) space group (Krklješ et al., 2013).

3.2 Electromechanical properties

The electromechanical response of the NBT single crystals was carefully tuned by changing the crystal surface with varying concentrations of KNN (K_0.5-Na_0.5NbO₃). Figure 8a shows that the piezoelectric coefficient d₃₃ falls gradually with increasing KNN concentration. The immaculate NBT single crystal shows a d₃₃ value of around 70 pC/N, which is still rather high up to 2 mol% KNN. A significant drop is found above 3 mol% with d₃₃ dropping to around 55, 35 and 15 pC/N at 3, 4 and 5 mol% KNN correspondingly. At higher concentrations, the piezoelectric response is nearly completely reduced, with d₃₃ values of ∼0.50 and 0.59 pC/N at 6 and 7 mol% KNN, respectively.

The large reduction of d₃₃ at 5 mol% is about 78.6% suppression as compared to pure NBT with a discernible piezoelectric response. In contrast, samples of 6 and 7 mol% show almost negligible d₃₃ values. Hence the 5 mol% composition provides a good compromise between the suppression of the response and the preservation of important electromechanical activity.

The related P-E hysteresis characteristics are shown in Figure 8b. All measurements were carried out at room temperature. The hysteresis behaviour is progressively modified with increasing KNN concentration. The 5 mol% KNN sample has a clearly detectable P–E response, whereas the 6 and 7 mol% samples exhibit a markedly decreased polarisation behaviour. The best combination of the d₃₃ and P–E data reveals that the most acceptable composition for the sensor integration in the concentrations studied is 5 mol% KNN. Hence, 5 mol% KNN modified NBT crystal is identified as an optimum composition for sensing application based on FSR. The NBT-based active element works by the piezoelectric effect, instead of a traditional resistive FSR that measures the force by a change in electrical resistance. The retention of d₃₃ at 5 mol% KNN provides a detectable electrical response to dynamic mechanical deformation and the significant decrease from the initial NBT value permits tailoring the electromechanical sensitivity for the intended sensing application. The 5 mol% KNN composition is a good compromise between a too high piezoelectric sensitivity at low KNN concentration and an almost total disappearance of the response at 6–7 mol%. Thus both data suggest 5 mol% KNN as a controlled electromechanical output for sensor optimization.

3.3 Mechanical properties of sensor

The hardness values are plotted against the applied load, and the graph is given in Figure 9. Due to the substantial crack start and chipping of the material at higher loads, the hardness test could not be carried out over 100 g. The hardness of the material is found to rise with increasing applied indentation load due to the softer superficial layers on the surface of the specimen. This observed increase in hardness with increasing stress is sometimes referred to as the reverse indentation size effect. Therefore, NBT can be utilized for device manufacturing (Diraco et al., 2013). Comparison of the harness before and after poling showed an increase of the harness by 3 times. It also showed an improvement in domain alignment and enhancement in mechanical rigidity (Rundo and Downey, 2019).

According to the graph (Figure 10), the value of n is 2.15 before poling and 1.07 after poling, representing the soft material before poling and the rigid material after poling. As a result, the NBT crystal prior to poling is considered flexible and simple to indent, resulting in increased plastic deformation and a large indentation size effect (ISE). However, following poling, the Meyer index decreased significantly, and the NBT single crystal became mechanically stronger, more resistant to deformation, and had stronger ferroelectric ordering. These NBT crystals are now more compact for device fabrication due to domain alignments following poling (Krklješ et al., 2013).

3.4 Sensor mechanism

The electromechanical response of the poled NBT single crystal was examined in terms of the d₃₃ value and the applied force. The charge created was calculated using the equation Q = F × d₃₃, where F is the applied force (N) and d₃₃ is the piezoelectric coefficient (pC/N), which results in the charge Q in pC. As illustrated in Figure 11, the produced charge increases with the increase of the applied force, which subsequently means that the sensor output can be improved by the increase in the applied mechanical force. The high d₃₃ value suggests the effective electromechanical conversion, and the greater Vickers hardness suggests the improvement of the mechanical stability under the applied stress. The combined electromechanical and mechanical features of the poled NBT single crystal indicate its potential for force and pressure sensing applications.

3.5 Implementation of FSR sensors in health monitoring systems

The embedded system (ESP32)-based electric control unit (ECU) is attached to the compact box, and then it is placed into the chair leg, as shown in Figure 12. The placement of the FSR sensors in the chair cushion based on anthropometric measurements of the Indian human body mass and structure.

The flexible rolling chair (Figure 12) can accommodate the proposed heterogeneous sensor system. The chair contains 6 FSR sensors embedded in the seat cushion to monitor pressure distribution. Furthermore, the ToF sensors are integrated on the backrest of the chair. It is situated at a suitable distance for a back-stinting attitude. As with ultrasonic sensor systems, the devices are not mounted on intact areas to limit the user’s pain. The sensor is in the sidewalls so that the pain is reduced and sinking into the backrest is avoided.

3.6 Calibration of the sensors

The six FSR sensors were calibrated separately to account for the inherent unit-to-unit fluctuation of their electrical responses. Each FSR was individually connected to a voltage-divider circuit with the 3.3 V supply of the ESP32 controller. One wire of the FSR was linked to the 3.3 V supply, and the other terminal was connected to a reference resistor of 10 kΩ and the ADC input of the ESP32. As shown in Figure 13, the opposite terminal of the reference resistor was connected to ground. The ESP32 ADC acquired the voltage across the reference resistor, which was used to determine the conductance of the corresponding FSR. The six FSR sensors were individually calibrated with regulated loads from 0.1 g to 1000 g (about 0.001–9.81 N). For each load applied, 125 samples were averaged to eliminate measurement noise and short-term changes. The recorded ADC voltage was transformed into the corresponding conductance by using the voltage-divider relation. The conductance was then plotted against the applied force to get the force–conductance characteristics of each individual FSR.

The calibration findings show a large difference from unit to unit of the six FSR sensors. A power-law relationship was employed to explain the force-dependent conductance response of each sensor (Martins et al., 2013). The individual calibration characteristics of the FSRs 1–6 are shown in Figures 14a–f. The FSR1, FSR4, and FSR5 mainly display linear force–conductance behavior (Figures 14a, d, e) Among all these sensors, the FSR5 presents the highest coefficient of determination with R² = 0.964, showing a good agreement between the fitted calibration model and the experimental data. Here, R² is the coefficient of determination and indicates the goodness of fit of the calibration relation. In contrast, as demonstrated in Figures 14b,c, FSR2 and FSR3 display hyperlinear responses, indicating that the larger the applied stressors, the more significant changes in conductance. This behavior shows that the sensitivity of these sensors is not constant over the range of applied force. Figure 14f shows the sublinear response of FSR6, which is related to its reduced baseline sensitivity (Rodríguez de Trujillo et al., 2018; Xu et al., 2023). These variances suggest that the six commercial FSR elements cannot be considered to have the same force-conductance properties.

The proposed rule-based classification algorithm compensates this unit-to-unit variability by using individual calibration features and sensor-specific conductance thresholds calculated for each FSR rather than one common absolute threshold for all six sensors. Each FSR is thus interpreted using its unique calibrated response profile. This method decreases the influence of the disparities in baseline sensitivity, slope, and non-linearity of the sensors. These calibrated conductance values are then utilized to determine the pressure distribution at the 6 sensor points, which is used for the posture categorization.

Therefore, the calibration and threshold selection specifically address these differences, whilst FSR1, FSR4, and FSR5 are linear, FSR2 and FSR3 are hyperlinear, and FSR6 is sublinear. The sensor-wise calibration technique allows the six FSRs to be integrated into a common rule-based posture-monitoring framework without the assumption of equal sensor properties (Rodríguez de Trujillo et al., 2018; Xu et al., 2023).

3.7 ToF calibration

The time-of-flight (ToF) sensors were calibrated by measuring distances to a set of specified reference places inside the sensing system. For each calibration point, a reference object was placed at a specified distance from the ToF sensor, and the distance measured by the sensor was recorded. To assess the repeatability, measurement stability, and the variability of the sensor output, 25 independent measurements were taken at each reference distance. The mean measured distance and its accompanying variation were then used to evaluate the sensor response. The calibration relationship was derived by comparing the measured distance (D_m) acquired from each ToF sensor with the reference or real distance (D_r) computed using a standard reference method. The obtained calibration data were plotted to get the relation between the sensor output and real distance. In Figure 15 the scatter plot shows 25 measurements for each reference-distance level so that the repeatability and the dispersion of the sensor readings can be assessed. The scatter observed suggests that the measurement variation grows gradually with sensing distance, indicating a deterioration in measurement consistency at longer working ranges.

The fitted a linear regression model to the calibration data, and this can be stated aswhere (Dm) is the measured distance, (Dr) is the reference distance, (m) is the slope and (c) is the intercept. The best-fit linear regression of the experimental data is shown as the red line in Figure 15. The slope close to unity shows high linearity in the investigated sensing range, since the change of the measured distance closely reflects the corresponding change of the reference distance. The calibration findings for both ToF sensors show slopes close to 1, showing a strong positive linear association between the measured and reference distances. The intercept values show a slight systematic offset of the sensor output, in particular for the shorter sensing-distance region. Such an offset can be corrected during calibration by applying the corresponding regression equation to the raw sensor values.

D m = mD r + c

The coefficient of determination (R² > 0.98) for both sensors shows a strong link between the measured and reference distances. This indicates that a linear calibration model can accurately represent the experimental data. The high values of (R^2) indicate that most of the variation in the observed distance can be explained by the corresponding reference distance. The calibrated ToF sensors thus offer excellent linearity and measurement consistency for the specified distance-monitoring application. The calibration technique also provides a correction relation that can be integrated into the sensing system to compensate for systematic measurement mistakes and increase the accuracy of the final distance calculation.

3.8 Posture classes

The proposed study included 10 healthy human volunteers with a job nature that requires them to spend most of their time sitting in a chair, such as office workers, students, and those with sedentary lifestyles. Before signing the consent forms, the participants would be informed about the experiment’s purpose and procedures. The proposed system is created to detect sitting posture and classify it as any predefined sitting posture. The system is trained to identify and categorise eight sitting postures, including upright sitting, forward slouch sitting, crossed legs (left leg over right and right leg over left), leaning left, leaning right, backward leaning, and sitting at the seat edge. The figure depicts the effect of the defined sitting posture (Figure 16).

3.9 FSR sensors

The FSR radar maps show a rapid visual representation of the posture-specific seat pressure distribution patterns for all six sensors. The upright posture has an almost round, symmetric radar map that indicates equal stress on the seat pan and promotes a balanced pelvic position. The forward slouch posture follows a similar pattern, but less distorted, in which the pressure is distributed less uniformly, resulting in forward bending of the trunk without loss of contact with the seat. On the other hand, the lean-back posture shows a clear collapse of the radar area along the central sensor axes, showing that the middle region of the seat has been emptied due to trunk movement backward to the backrest. The lateral postures are highly skewed; left-leaning postures suggest significant sensor activation on the left, and vice versa. The seat-edge position is associated with an extremely skewed radar profile, with the anterior sensors robustly activated and the posterior sensors very weakly activated, indicating pelvic forward translation with no back support. Crossed-leg postures correspond to the most irregular radar profiles, which are extremely asymmetric and erratic in sensor activation. The left-over-right and right-over-left postures exhibit similar dominance patterns, demonstrating the pressure redistributions caused by crossing the legs and shifting the pelvis. In conclusion, these radar plots show that each sitting posture has a distinct pressure pattern, which supports the feature extraction approach and the robustness of the rule-based posture classification system (Figure 17).

3.10 TOF sensors

The high open percentage of the upper ToF sensor for slouch (91.6%) and seat edge (100%) indicates that the upper back in these circumstances is largely detached from the backrest. On the other hand, upright and cross-leg open fractions are small, indicating ongoing linkages to the backrest. The lower ToF sensor shows a high response only for the seat-edge posture (89.2%), suggesting less contact with the lower back when users sit forward. Further, low activity corresponds to lateral leaning and cross-legged postures, showing that ToF sensors are hardly significant to postural variations dependent on the legs. The ToF open percentage in Table 1, offer a quantification of the sensor activity in a posture-specific way and are compatible with the visual trends shown in the Figure 18.

TABLE 1. ToF open percentage by posture.

The obtained FSR and ToF sensor data were classified into eight different sitting postures using the rule-based classifier. The ToF measurements gave additional posture-differentiating information from the upper and bottom sensor responses. As presented in the calibration data, the forward-slouch posture gave a clear response from the upper ToF sensor (91.6%), whereas the bottom ToF was closed (0%). The seat-edge posture showed the strongest response for both sensors, 100% upper-ToF opening and 89.2% bottom-ToF opening. The upright posture, however, demonstrated insignificant ToF responses (0.1% upper and 0% bottom), but the lean-back posture elicited just 1.2% and 0.2%, respectively. The left- and right-tilted postures resulted in comparatively low upper-ToF responses of 7.6% and 4.5%, respectively, without bottom-ToF opening. ToF responses were also minimal in both cross-legged postures.

Figure 19 shows the confusion matrix of the classification performance of the proposed posture-monitoring system. The system achieved an overall classification accuracy of 91.3% for the eight sitting positions tested. The best classification accuracy was obtained with the upright position (100%), showing that it was well distinguishable from the other postures. Recognition accuracy for the lean-back position was likewise good (97.7%). The forward-slouch posture was identified with 91.6% accuracy, while about 8.3% of the samples were misclassified as upright. The unusually high upper-to-FrF response of 91.6% seen for forward slouch gives an extra discriminative feature for this posture.

The left-leaning posture and the right-leaning posture achieved classification accuracies of 86.7% and 85.2%, respectively. The right-over-left cross-legged posture was correctly classified in 92.7% of the cases, with only slight misunderstanding with the upright posture. The left-over-right cross-legged posture also achieved 92.7% accuracy. However, the seat-edge posture produced the lowest classification accuracy, 77.9%, despite having the highest ToF responses. The combined FSR pressure distribution and positional parameters for this posture are similar, which may be responsible for this substantially worse categorization performance.

Overall, the results reveal the complementary information of the FSR and ToF sensing approach for posture recognition. The FSR sensors capture the pressure distribution generated by the user’s body, while the ToF sensors provide additional information about the user’s body position and distance from the detecting regions. The inclusion of these variables in the rule-based classifier allows for a valid differentiation of the eight sitting postures tested with an overall accuracy of 91.3%.

The comparison shows (Table 2) that the suggested system achieves competitive posture classification performance with a reasonably compact sensing setup. Existing chair-based systems have employed dense arrays of pressure sensors, FSR arrays, piezoelectric pressure sensors and hybrid pressure-distance systems. The number of categorised postures reported varies widely and the accuracy reported is affected by the amount of sensing elements, sensing modality, participant population, similarity of postures and the classification algorithm. The proposed approach classifies eight sitting postures with 91.3% overall accuracy with 10 participants. Compared to traditional FSR-only systems, the suggested design involves six FSR sensors integrated with ToF sensing and a piezoelectric sensing element using NBT single-crystal. The FSR sensors give local force/pressure information while the ToF sensors give supplementary non-contact positioning information. This combination is useful to distinguish between postures with comparable pressure distribution but different body postures. The proposed system likewise achieves equivalent performance (92%) to the reported FSR + infrared-distance system, but uses ToF sensing as the supplemental distance-sensing modality. The present work therefore indicates that a relatively small number of sensing elements can provide useful posture discrimination when merged with complementary sensing information and a rule-based classification technique. The accuracy of 91.3% in eight positions is of special relevance as the suggested system is to be a practical, low-complexity posture monitoring platform, rather than depending on a high-density pressure mat.

TABLE 2. Comparison of representative chair-based posture-monitoring systems.

4 Conclusion

The integration of electrically poled Na_0.5Bi_0.5TiO₃ (NBT) single crystals with FSR and ToF sensing technologies resulted in the successful development of a novel, compact, and versatile smart posture-monitoring system. The NBT single crystals were synthesised by a slow bottom-cooling flux-growth method. The effectiveness of the adopted crystal-growth methodology was confirmed by the X-ray diffraction analysis, which revealed a highly crystalline rhombohedral perovskite phase without any detectable impurity phases. The electrical poling of NBT crystals improved their electromechanical activity and ferroelectric domain alignment, which led to superior electromechanical coupling and enhanced piezoelectric charge generation capabilities. The systematic KNN modification revealed that the piezoelectric coefficient (d₃₃) can be effectively tuned by controlling the concentration of KNN, and the optimised condition with a significantly reduced but measurable d₃₃ response suitable for the proposed FSR sensing application was found to be 5 mol% KNN. Further support of the concentration dependent modification of the polarisation behaviour was given by the related P-E characteristics. The microhardness tests demonstrated an increase of mechanical stiffness, indentation resistance, structural compactness and durability under mechanical loading after electrical poling. The Vickers hardness and Meyer index were also raised. The force-dependent rise of the generated piezoelectric charge validated the efficient mechanical-to-electrical energy transduction and the effective force sensitivity of the NBT-based sensing device. In a small sensing system we improved posture discrimination by combining calibrated FSR sensors with ToF based positional sensing, by complementary acquisition of localised pressure distribution and non-contact geometry information. The suggested method attained an amazing total accuracy of 91.3% in classifying eight different sitting positions, which demonstrated the practical efficacy of the framework that integrates sensing with rule-based classification. To summarise, the proposed compact FSR-ToF posture-monitoring platform shows great promise for next-gen force, pressure, and ergonomic posture-monitoring applications due to its mechanical robustness, low-complexity sensing, tunable piezoelectric functionality and multimodal signal acquisition.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Bibhas Kumar Bhunia, Parul University, India

Reviewed by: Jiahe Li, Sichuan University, China

Tejas Shelgaonkar, KDM Engineers India Pvt Ltd. Hyderabad, India