Cambridge University: Isotropic and Omnidirectional Ultra sensitive Strain Sensing and Deep Learning Assisted Direction Recognition in Biomimetic Stretchable Devices

Flexible and stretchable physical sensors with omnidirectional sensing capabilities are of great significance in addressing complex, variable, and dynamic real-world scenarios in medical monitoring, human motion detection, and human-machine interfaces. In order to quantify vibration and deformation stimuli, stretchable strain sensors play a crucial role in wearable electronic products and electronic skins, with the advantages of high flexibility, simplicity, and consistency. Significant achievements have been made in the development of stretchable strain sensors, with a focus on improving their sensitivity, stretchability, durability, hysteresis, and detection limit by utilizing novel nanomaterials and micro/nanostructures. It is worth noting that excellent sensitivity is required within a small strain range to enable the sensor to detect tiny biophysical signals such as pulse waves and throat vibrations. However, due to their inherent characteristics such as large aspect ratio structures and unidirectionally distributed sensing materials, most high-performance stretchable strain sensors are limited by their ability to only convert uniaxial strain into electrical signals, hindering their application in multi axial strain environments. Therefore, there is an urgent need to develop more complex strain sensor systems that can effectively perceive complex information containing strain from various directions.

Recently, in order to detect more complex multi axis strain conditions, omnidirectional strain sensing technology has been developed mainly through two strategies: single sensor and multi-sensor systems. In single sensor schemes, some isotropic and omnidirectional flexible strain sensors are designed using curved microgrooves arranged around a circle and incorporating chiral stretchable metamaterials into the substrate. Although these sensors can detect strain from multiple directions with high sensitivity, they cannot determine the specific direction of strain with a single sensor, thus requiring additional sensor arrays. Previous attempts to implement directional strain sensing were based on multi-sensor system methods, typically involving two or three anisotropic strain sensors positioned at specific angles, as well as customized algorithms for calculating strain intensity and direction based on the signal difference between each sensor. This is fundamentally different from the single sensor method proposed in this work. In addition, due to the lack of isotropic properties, multi-sensor system methods are more complex when dealing with multi-directional strain (i.e. applying the same strain in different directions). Technically speaking, achieving isotropic and omnidirectional strain sensing and direction recognition within a single sensor is highly challenging because the fundamental principles are fundamentally opposite. Isotropic sensing requires a uniform platform to output the same response, while direction recognition relies on differences in signals. Therefore, considering the expected simplicity and efficiency of sensors, there is an urgent need for a feasible strategy that combines these two features for various practical applications.

Source: Sensor Expert Network