In recent years, e-skin, as an emerging flexible sensor device with its skin-like structure and function, has been receiving extensive attention from scholars for its important applications in the fields of healthcare, haptic feedback bionic electronics, and robotics. Especially in the role of robot tactile perception is particularly critical, the e-skin not only provides a bridge for human-machine-object interaction, but also through its unique tactile channel, endows the robot with a richer perceptual ability, and shows great potential in improving the comfort and safety of human-robot interaction. In order to accurately sense and respond in complex environments, electronic skins are required to have characteristics such as ultra-wide range, high linearity, and high consistency. The commonly adopted technology path is printed electronics, which has the advantages of low printing cost, short cycle time, mass production and high adaptability, providing new possibilities for the manufacture of electronic skin and becoming the preferred technology for the large-scale manufacture of electronic skin.

Recently, the Robot Technology and System Center of the Chongqing Institute of Green and Intelligent Technology, Chinese Academy of Sciences, published a paper titled "Leather-Based Printed Tactile Sensor Array for Robotic Interactive Skin" in the journal Nano Energy. This study has developed a novel silk-screen printing technique for leather-based microstructure surfaces, enabling the fabrication of an arrayed electronic skin with vertically gradient conductive fiber networks. This unique multilayer hierarchical conductive fiber network force-sensitive structure allows the sensor to maintain high linearity (99.30%) and high uniformity (99.98%) within a broad pressure range (0-4.5 MPa). Additionally, it exhibits rapid response time (10 ms), fast recovery time (<10 ms), excellent stability (~7500 cycles), and stable dynamic response performance. At the same time, its excellent flexibility allows the haptic leather device to be better conformally mounted on the robot's body surface. Additionally, using Tet-Net Convolutional Neural Network as the backbone network, combined with depthwise separable convolution, multi-scale modules, asymmetric convolutions, etc., we established a Contact Object Recognition Residual Network (COR-Net) and an Interaction Gesture Recognition Residual Network (IGR-Net) based on the attention mechanism to successfully recognize the hardness material of different objects for collision warning and judge human interaction intentions, with accuracy rates of 95.00% and 98.48%. Robots can complete object grasping, handling, and other operations according to human interaction intentions. The study confirms that e-skin technology can greatly enhance the safety and naturalness of robotic human-robot interactions.

Bingxue Zhang, a graduate student at the Center for Robotics and Systems, is the first author of the paper, and Professor Dapeng Wei is the corresponding author. The results of this research demonstrated the great application potential of flexible printed haptic leather devices in the field of human-computer interaction, which was supported by the projects of Chongqing Science and Technology Bureau, the Department of Science and Technology of the Tibet Autonomous Region, and the Talent Program of Chongqing Municipality.

Links to related papers： https://doi.org/10.1016/j.nanoen.2024.110379

E-Skin System