Human Skin Inspired Tech Redefines Robotics and Wearable Devices

image-20.jpg

Human skin is a prodigious organ, capable of sensing pressure, temperature, texture, and more with incredible sensitivity. A 2025 study in Soft Science by Jianfeng Sun, Chenyu Zhang, and colleagues explores how skin-inspired neuromorphic sensors can replicate these abilities, opening new possibilities for robotics, healthcare, and wearable devices. These sensors are paired with neuromorphic computing, a technology that mimics the brain’s efficient processing, to create intelligent systems that can perceive and respond to their environments in real time while using very little energy. Unlike standard computing methods that convert sensory data into digital signals, leading to massive data processing needs, high energy costs, and communication delays, neuromorphic systems process data right where it’s sensed, much like the human brain, which operates on just 20 watts compared to a computer’s 250 watts for similar tasks. This advancement could benefit several public companies across multiple sectors, from robotics to medical diagnostics.

The study details how skin-inspired sensors use flexible and stretchable materials, such as a silicone like material called polydimethylsiloxane (PDMS), conductive polymers, and graphene, to detect a range of sensations. Inclusive is touch, warmth, and moisture. These sensors are designed to act like human skin, with layers that mimic its structure to achieve high sensitivity. For example, a material called laser-induced graphene (LIG) is praised for its ability to be fine-tuned, making it perfect for creating sensors that can do numerous things at once. The sensors work in three main steps. First, they react to physical sensations using designs that copy nature. Second, they turn these sensations into electrical signals, such as when a sensor detects pressure by changing its electrical resistance as it gets squeezed, or by altering its ability to hold an electrical charge when pressed. Third, brain-inspired algorithms, rooted in neuromorphic computing, process these signals right at the sensor level, reducing the need for a big central computer and allowing for quick reaction times. This local processing, enabled by devices that act much like brain synapses, lets the sensors perform tasks like adjusting a robot’s grip based on how much pressure it feels or noticing temperature changes on a wearable device.

A key part of the research involves multimodal perception, the combination of information from various sensors for a more complete understanding of the surroundings. To illustrate, integrating touch data with visual data can enhance object recognition by providing details about texture and how the material feels to the touch that a camera alone cannot capture. The study details how convolutional neural networks, a form of artificial intelligence, analyze visual information, while pressure patterns over time process touch information. Subsequently, these data streams are merged to improve the system’s accuracy, mirroring how human skin uses different senses to perceive the world.

The paper details diverse skin inspired sensors designed to perceive temperature pressure and multiple stimuli concurrently. Temperature sensors gauge warmth by assessing alterations in electrical resistance mirroring the function of a thermometer with precision adequate for wearable technology. Pressure sensors register touch and movement employing different principles such as resistance variation under pressure or electrical charge displacement upon squeezing. A flexible pressure sensor crafted from polyimide nanofiber films exemplifies this capability sensing the delicate touch of a feather yet also enduring substantial pressure from a firm grasp. Furthermore certain sensors integrate multiple sensing modalities such as temperature pressure and noncontact proximity detection. A prominent illustration is a flat printed sensor sheet incorporating a unit grid sixteen for pressure, four for temperature and one for proximity allowing simultaneous capture of these sensory inputs.

At the heart of these sensors’ intelligence is neuromorphic computing, which uses devices like memristors and transistors to once again replicate the brain’s neural networks. Memristors act like artificial synapses by remembering electrical activity, allowing them to process data right where it’s sensed, a process called compute-in-memory. This eliminates the need to shuttle data between separate memory and processing units, saving energy and speeding things up. Memristors come in various forms, using materials like metal oxides or organic polymers, and some are made from ultra-thin 2D materials like hexagonal boron nitride, which offer fast switching and low energy use. They’ve been used in tasks like motion detection and robot navigation, showing their versatility. Transistors, another core component, can also mimic synaptic behavior by adjusting electrical flow, with designs like bottom-gated, side-gated, floating gate, or top-gated structures offering flexibility for different applications. Consider a device called a neuro-inspired artificial tactile neuron (NeuroMAT) that used transistors to control a robotic hand’s grip with very little power; similarly, a memristor paired with a pressure sensor recognized letter patterns with 98 percent accuracy using a brain-like system called a spiking neural network (SNN).

These brain-like systems, or SNNs, process data using spikes, much like neurons in the brain, making them far more energy-efficient than standard artificial intelligence systems. A learning method called spike-timing-dependent plasticity (STDP) adjusts the strength of connections between artificial neurons based on the timing of their spikes, imitating how the brain itself acquires knowledge. While SNNs are harder to train than standard artificial neural networks due to their intricate, spike-based nature, researchers are improving their accuracy by borrowing techniques from traditional AI, achieving up to 99.50 percent success in tasks like recognizing images. Frameworks like SpikingJelly help make SNNs more scalable, supporting their use in real-world applications, such as processing data on small devices.

A significant advancement in the study is the integration of sensors and neuromorphic processing onto a single system-on-chip (SOC). This on-chip design creates a compact unit for sensing, processing, and decision-making. Consequently, energy consumption is lower, and real-time tasks become feasible. For instance, the system can recognize patterns or make decisions immediately. The operational flow begins with sensors gathering environmental signals, such as temperature or pressure. Subsequently, neuromorphic devices like memristors or transistors process this input. The data is then converted for neural network analysis, and brain-like algorithms generate rapid responses.

Applications of this technology span acrossmultiple fields. In electronic skin (e-skin), sensors allow robots to interact safely with humans by detecting pressure and temperature, as shown in a multifunctional e-skin system that can even repair itself. Smart prosthetics gain better sensory feedback, with systems using muscle signals to control bionic hands with precision. Wearable devices, such as a flexible sweat-sensing platform, can monitor health signals like sweat composition in real time, potentially detecting dehydration or blood sugar levels without needles. Robotic skin, equipped with touch-sensing arrays, helps robots adjust their actions based on what they feel, improving their ability to avoid obstacles and interact with objects. Multimodal perception systems combine touch and sound data, enabling complex tasks like syncing touch with voice commands for smarter control.

Several public companies could benefit from this technology. In robotics, ABB (NYSE ABB) could integrate these neuromorphic sensors into industrial robots, enhancing their ability to handle delicate objects or navigate complex environments through improved tactile feedback and real-time processing. Boston Dynamics, often partnered with public firms like Hyundai (KRX 005380), might use this technology to advance humanoid robots like Atlas, enabling more human-like interactions with their surroundings through brain-like sensory processing. In healthcare, Medtronic (NYSE MDT) could incorporate these sensors into smart prosthetics, allowing amputees to feel pressure and temperature through artificial limbs, with neuromorphic chips enabling faster, more natural responses. Wearable device companies like Fitbit, owned by Alphabet (NASDAQ GOOG), could develop advanced health monitoring patches that track temperature and pressure to catch early signs of issues like infections or pressure sores, leveraging neuromorphic efficiency for longer battery life.

Medical diagnostics firms also stand to gain. Becton Dickinson and Company (NYSE BDX) might explore these sensors for advanced diagnostic tools, using their multimodal capabilities and neuromorphic processing to monitor patient vitals more effectively in real time. Sysmex Corporation (TYO 6869), known for blood analysis machines, could use the sensors’ high sensitivity and brain-like algorithms to create non-invasive monitoring systems for chronic conditions, expanding their diagnostic offerings. In the technology sector, Intel (NASDAQ INTC) could apply its neuromorphic chips, like Loihi, to power these sensor systems, enhancing their ability to process sensory data efficiently, while NVIDIA (NASDAQ NVDA) might adapt its GPUs to support SNN training, enabling real-time applications in robotics and healthcare. Additionally, companies like 3M (NYSE MMM), which makes advanced materials and wearable health monitors, could use these sensors to create smart bandages or patches for continuous patient monitoring, with neuromorphic integration ensuring low power use and high accuracy.

Despite their enormous potential, the sensors do have their impediments. Standard manufacturing methods struggle with large-scale production, and integrating new devices like memristors into big systems is tricky due to issues with fabrication and reliability. There’s also interference between different sensing types, high energy use in large systems, and difficulties in making them widely available. Manufacturing costs are high, and production is complex, making it hard to scale up for broad use. For e-skin and robotic skin, long-term comfort needs improvement, especially in terms of softness and breathability. Large sensor arrays often suffer from crosstalk, where signals interfere with each other, and performance can degrade over time. The study suggests subsequent research should focus on developing new materials, like flexible conductive polymers and tiny nanomaterials, to create large, multifunctional, and intelligent sensor arrays with neuromorphic integration for better performance. Advances in 3D printing and nanotechnology could help make production easier, while combining different sensor types might reduce interference. Improving neuromorphic SOC technologies, through methods like stacking components in layers or using laminated electronics, could make these systems more energy-efficient, compact, and self-powered, paving the way for practical use in the real world.

The continued advancement of this technology promises to usher in an era of more intelligent and reactive devices. This evolution brings us closer to creating systems that not only emulate human senses but also provide sustainable, energy-efficient solutions. From tactile robots to real-time health-monitoring wearables, skin-inspired neuromorphic sensors are poised to reshape our interaction with technology, generating profound impacts across diverse industries.

Citation
Sun, J.; Zhang, C.; Yang, C.; Ren, Y.; Ma, B.; Yuan, W.; Ye, T. Human skin-inspired neuromorphic sensors. Soft Sci. 2025, 5, 18. http://dx.doi.org/10.20517/ss.2024.77

NeuromorphicCore

Related Articles

PHP Code Snippets Powered By : XYZScripts.com
★

Track This Company

Create a free account to add companies to your watchlist and get notified when patents, funding, or key developments change.