Scientists have recently engineered an artificial flexible sensory nerve capable of neural coding, tactile sensing, and performing synaptic processing functions. Interestingly, this device does not depend on algorithms or computer resources. The study is available in Advanced sciences.
Study: A Flexible Artificial Sensory Nerve Activated by Nanoparticle Assembled Synaptic Devices for Neuromorphic Touch Recognition. Image credit: Christoph Burgstedt/Shutterstock.com
Tactile perception and recognition in humans
In humans, tactile recognition and perception have been associated with determining the strength and dynamics of sensory stimuli, which are submitted to the skin via touch (active or passive). External stimuli or touch are perceived by sensory receptors, which are present on the skin, and are encoded as neural spikes.
These spikes are handled by neurons and synapses, through various processes, such as amplification, adaptation, and memory, and then transmitted to the cerebral cortex. The cerebral cortex is involved in high-level functions including identification, classification, and perceptual learning.
Due to tactile perception, humans can perform delicate input, distinguish various textures, and identify different objects based on their tactile characteristics and patterns.
Neuromorphic electronics
Scientists believe that integrating tactile functions into prosthetics, robotics and other interactive systems would significantly improve an individual’s cognitive skills when encountering unstructured environments or maneuvering unidentified objects. Advances in neuromorphic electronics have helped enormously in the design of artificial tactile sensory systems.
The scientists said an artificial afferent nerve, which includes organic synaptic transistors and networks of pressure sensors, is able to distinguish Braille characters. This is based on optoelectronic afferent nerve output (sensor signals) in the form of optical tips which are processed via a software algorithm and detect touch and recognize handwriting.
Some techniques used to process the touch data obtained from the synaptic device are machine learning algorithms and deep learning. Consequently, these synaptic electronics require additional calculation processors, and the tactile recognitions are not carried out in real time.
Nano-Based Neuromorphic Touch Recognition Device – A New Study
In a new study, scientists have developed a flexible artificial sensory nerve that can perform neuromorphic touch recognition in real time. This touch sensor featured high sensitivity and linear response for touch sensing performance. A flexible synaptic transistor has been associated with sensory processing based on neural sensitivity and sensory memory.
Scientists designed the synaptic device using a simple technique associated with nanoparticle (NP) self-assembly that can produce continuous and uniform films of NPs on arbitrary substrates. Above all, the touch data was processed directly by the device without requiring external algorithms or computing resources. Additionally, the synaptic device also did not require a memory wipe or state reset operation.
In this study, the synaptic device is an ion-gate transistor composed of interdigital electrodes, a chitosan-based electrolyte placed on a flexible polyimide substrate, and a self-assembled NP channel. The researchers said the NP interfacial self-assembly technique is an efficient and inexpensive fabrication process. In this study, researchers used colloidal amphiphilic zinc oxide NPs.
The touch sensor was designed based on the lamination of the pressure-sensitive layer, composed of carbon nanotubes (CNTs) and room temperature vulcanizing latex (RTV) on a polyimide film coated with gold electrodes. The interdigital electrodes have been bridged by a conductive sensing layer.
The authors stated that the flexible artificial sensory nerve processed stimuli similar to signal transmission and had processing characteristics associated with the biological tactile sensory system. The authors of this study pointed out that their simplified design of the artificial sensory system dramatically elevated the neuromorphic perception capabilities of the device. The output of the device, i.e. synaptic weight and peak pulse count, is classified for real-time intelligent touch recognition during robotic manipulation and touch interaction.
Hardness is one of the most important characteristics of an object that humans learn through touch. In this study, the new device’s touch recognition capabilities were assessed by its ability to identify the hardness of a material during a robotic slow grip and quickly release it.
The force of contact between the object and the finger during a touch was assessed initially by the touch sensor followed by the synaptic device that was attached to the robotic fingers. The tapping pattern was studied to assess the touch interaction capability of the device.
The authors revealed that when gripping a porous material with low hardness, the postsynaptic current (PSC) showed inconsistencies. This may be due to the release of compressive pressure on the material.
Conclusion
This study provides a guideline for the development of a neuromorphic sensory system that possesses human-like cognitive functions. As the newly developed synaptic device and touch sensor are both flexible, they can be integrated into a wide range of wearable electronics and robotics.
One of the major advantages of this device is that the neuromorphic touch recognition is directly obtained at the device level without depending on an algorithm or a computer resource. Going forward, the authors will focus on miniaturization and optimization of the S-coding circuit to develop flexible electronics with tunable synaptic plasticity and long-term memory in the synaptic device.
Reference
Jiang, C. et al. (2022) A flexible artificial sensory nerve activated by nanoparticle-assembled synaptic devices for neuromorphic touch recognition. Advanced science. https://doi.org/10.1002/advs.202106124
#Artificial #sensory #nerve #based #nanoparticles