Smart Human-machine Interface Research Articles

Flexible Conformally Bioadhesive MXene Hydrogel Electronics for Machine Learning-Facilitated Human-Interactive Sensing.

Wearable epidermic electronics assembled from conductive hydrogels are attracting various research attention for their seamless integration with human body for conformally real-time health monitoring, clinical diagnostics and medical treatment, and human-interactive sensing. Nevertheless, it remains a tremendous challenge to simultaneously achieve conformally bioadhesive epidermic electronics with remarkable self-adhesiveness, reliable ultraviolet (UV) protection ability, and admirable sensing performance for high-fidelity epidermal electrophysiological signals monitoring, along with timely photothermal therapeutic performances after medical diagnostic sensing, as well as efficient antibacterial activity and reliable hemostatic effect for potential medical therapy. Herein, a conformally bioadhesive hydrogel-based epidermic sensor, featuring superior self-adhesiveness and excellent UV-protection performance, is developed by dexterously assembling conducting MXene nanosheets network with biological hydrogel polymer network for conformally stably attaching onto human skin for high-quality recording of various epidermal electrophysiological signals with high signal-to-noise ratios (SNR) and low interfacial impedance for intelligent medical diagnosis and smart human-machine interface. Moreover, a smart sign language gesture recognition platform based on collected electromyogram (EMG) signals is designed for hassle-free communication with hearing-impaired people with the help of advanced machine learning algorithms. Meanwhile, the bioadhesive MXene hydrogel possesses reliable antibacterial capability, excellent biocompatibility, and effective hemostasis properties for promising bacterial-infected wound bleeding.

Intercalation Engineering of 2D Materials at Macroscale for Smart Human–Machine Interface and Double‐Layer to Faradaic Charge Storage for Ions Separation

Abstract2D materials have attracted considerable attention in the past decade for their superlative physical properties. Lightweight foam from 2D materials can reduce the use of raw materials, save energy in processing the material and downregulate the carbon emission. Self‐assembling of graphene nanosheets results in 3D porous lightweight foam resembling hydrogels, aerogels, and xerogels. Hierarchical graphene architecture provides an uninterrupted path to electrons and phonon transport conforms to eligibility for developing sensor, energy storage, and energy conversion devices. Artificial intelligence (AI) originates from human–machine interaction and physiological signal reception which requires a large sensing range that illustrates to healthcare surveillance with wearable devices. Spongy architectures with flexibility and reversible compressibility act as a good candidate for pressure sensing for a wide range of real‐time human health monitoring systems by using artificial intelligence. In this review, the new perspective of emerging 2D materials (such as MXene, and borophene) along with the investigation of the graphene foam structure is demonstrated. Electrosorption of salt‐ions under foam electrode are reviewed in the context of electrochemical stability recycling ability, and efficiency of adsorption‐desorption. Beyond these, remaining challenges are depicted for emerging 2D‐materials of interest for health monitoring sensors and flexible supercapacitors for promising research directions.

REAL TIME AGE AND GENDER CLASSIFIER

Gender is still a prominent element of our personalities. It also plays an important role in our social lives. Artificial intelligence age forecasts have applications in a variety of disciplines, including smart human-machine interface development, health, cosmetics, and electronic commerce, among others. The ability to estimate people's sex and age from their face photographs is a current and active research topic. The researchers proposed a number of solutions to the problem, but the criteria and actual results are still insufficient. This paper proposes a statistical pattern recognition strategy to solve this challenge. In the proposed method, a Deep Learning technique called Convolutional Neural Network (ConvNet / CNN) is employed to extract features. CNN takes input images and assigns a value to distinct characteristics / elements of the image (learnable weights and biases) and can distinguish between them. Other classification techniques require far more pre-processing than ConvNet. While the filters are created by hand using simple methods, ConvNets can be trained to learn these filters and features. Face photos of individuals were trained with convolutional neural networks in this study, and age and sex were accurately predicted with a high rate of success. There are around 20,000 photos with age and gender. Poses, facial expressions, lighting, occlusion, and resolution are all represented in the photos.

Printed, Wireless, Soft Bioelectronics and Deep Learning Algorithm for Smart Human–Machine Interfaces

Recent advances in flexible materials and wearable electronics offer a noninvasive, high-fidelity recording of biopotentials for portable healthcare, disease diagnosis, and machine interfaces. Current device-manufacturing methods, however, still heavily rely on the conventional cleanroom microfabrication that requires expensive, time-consuming, and complicated processes. Here, we introduce an additive nanomanufacturing technology that explores a contactless direct printing of aerosol nanomaterials and polymers to fabricate stretchable sensors and multilayered wearable electronics. Computational and experimental studies prove the mechanical flexibility and reliability of soft electronics, considering direct mounting to the deformable human skin with a curvilinear surface. The dry, skin-conformal graphene biosensor, without the use of conductive gels and aggressive tapes, offers an enhanced biopotential recording on the skin and multiple uses (over ten times) with consistent measurement of electromyograms. The combination of soft bioelectronics and deep learning algorithm allows classifying six classes of muscle activities with an accuracy of over 97%, which enables wireless, real-time, continuous control of external machines such as a robotic hand and a robotic arm. Collectively, the comprehensive study of nanomaterials, flexible mechanics, system integration, and machine learning shows the potential of the printed bioelectronics for portable, smart, and persistent human-machine interfaces.

A PVDF/Au/PEN Multifunctional Flexible Human-Machine Interface for Multidimensional Sensing and Energy Harvesting for the Internet of Things

In the picture of the Internet of Things (IoT), smart human-machine interfaces (HMIs) play a significant role. Desired attributes for smart HMIs include multidimensional sensing, flexibility with stable responsiveness, and a long operational lifetime. To this end, in this article, we present a polyvinylidene fluoride (PVDF)/Au/polyethylene naphthalate (PEN) sandwiched architecture. Piezoelectric property of the PVDF is utilized for dynamic force sensing and touch-induced energy harvesting. A mutual capacitive based technique is employed for two-dimensional (2D) position recognition, hover distance measurement and curvature self-detection, as well as helping in compensating the static force detection shortage of the piezoelectric sensing layer. The presented HMI system achieved an overall force sensitivity at 0.07 N and capacitive resolution at 0.004 pF; and a force responsivity at 2.36 V/N in the pressure range of 1 N-30 N. In each touch event, the maximum capacitance change and the electric energy harvested in a stored capacitor are averaged at 0.016 pF and 0.038 nJ, respectively. Furthermore, interdigitated electrodes are implemented for self-detecting the device curvature. With pre-correlating of the force-voltage response under different bending degrees, a stable force detection response is realized when the shape of the device changes. The work presented in this article advances the field of human-machine interactivities.

Human-robot coexistence and interaction in open industrial cells

Abstract Recent research results on human–robot interaction and collaborative robotics are leaving behind the traditional paradigm of robots living in a separated space inside safety cages, allowing humans and robot to work together for completing an increasing number of complex industrial tasks. In this context, safety of the human operator is a main concern. In this paper, we present a framework for ensuring human safety in a robotic cell that allows human–robot coexistence and dependable interaction. The framework is based on a layered control architecture that exploits an effective algorithm for online monitoring of relative human–robot distance using depth sensors. This method allows to modify in real time the robot behavior depending on the user position, without limiting the operative robot workspace in a too conservative way. In order to guarantee redundancy and diversity at the safety level, additional certified laser scanners monitor human–robot proximity in the cell and safe communication protocols and logical units are used for the smooth integration with an industrial software for safe low-level robot control. The implemented concept includes a smart human-machine interface to support in-process collaborative activities and for a contactless interaction with gesture recognition of operator commands. Coexistence and interaction are illustrated and tested in an industrial cell, in which a robot moves a tool that measures the quality of a polished metallic part while the operator performs a close evaluation of the same workpiece.