New Robotic Skin Enables Machines to Feel Like Humans

A team of engineers from the University of Cambridge and University College London collaborated to create robotic skin that can equip nearly any machine with the ability to sense. Their unique approach leverages multi-modal sensors, hydrogel materials, and machine learning to create a new level of tactile feedback that has the potential to create prosthetics that can feel and much more. Here’s what you need to know.

Why Touch Matters in Robotics

The five senses that enable humans to navigate the world are also crucial for creating robots that can interact with humans and their surroundings in meaningful ways. It’s common to see robots integrating more sensory inputs as part of their added capabilities.

However, the sense of touch is just now getting the attention it deserves. From industrial robots that can tell when they bump into another worker, to agricultural robots that can feel if a fruit is ripe, touch sensing has now become a vital aspect of the technology.

How Human Skin Inspires Robotic Design

When it comes to touch, your skin is a marvel. It can accurately gauge multiple types of input, from heat to pressure and more, in milliseconds. Additionally, this can happen over the entire body or in particular areas, with your brain able to differentiate the origins and type of touch. All of these capabilities come from a flexible surface that can heal itself.

Giving Robots the Ability to Feel

Recognizing that human skin sensitivity is ideal for many applications, engineers have attempted to create robotic skins on multiple occasions. The primary way that they approached this task was to embed several sensors into a flexible material. The sensors could detect pressure, temperature, and other vital data.

Problems with Robotic Skin

This strategy had some major drawbacks. For one, every sensor added to the rigidity of the skin. Also, creating these sensors and installing them into the device was a costly and time-consuming process. These devices were very expensive to create and were, by all accounts, fragile.

Every rigid sensor added another area that could be damaged in operations. The sensor placement also played a role in what shapes and designs the surface needed to be, further limiting applications. Additionally, many of the sensors had blind spots where they weren’t located or where other sensors would need to be fitted. On top of these problems, each new sensor type added to the interference created by the device, reducing accuracy.

Inside the Robotic Skin Study

The Multimodal information structuring with single-layer soft skins and high-density electrical impedance tomography study¹ introduces a novel type of robotic skin that’s capable of detecting multiple types of touch from a single material design. The device can be fitted to any shape and can be added to existing robotic hands like a glove, providing a new layer of interactivity and sensitivity to the device.

Source – University of Cambridge

Multi-Modal Sensing

The first step in creating the enhanced robotic skin was to find out how to reduce the number of sensors needed. To accomplish this task, the engineers turned towards multimodal sensors. These devices are capable of registering different types of signal input and separating the cause of each without interference. As part of this approach, the team developed a purpose-built system that supported 6 types of multimodal stimuli, including heating, touch, pressure, damage, and more.

Why Hydrogel Is Key to Robotic Skin

A conductive gelatine-based hydrogel was then cast into the shape of an adult hand. This material was selected due to its conductivity, soft feel, and flexibility. Additionally, it can be melted down and formed into whatever shape is needed, adding to the usability and sustainability of these devices.

Electrode Configurations

The team spent significant time testing different electrode layouts. Their goal was to find a setup that didn’t interfere with the robotics skin’s capabilities but provided accurate measurements to the sensors. The team eventually settled on a design that placed 32 electrodes at the wrist.

Electrical Impedance Tomography

The engineers fed these sensors data utilizing electrical impedance tomography techniques. Keenly, the conductivity of the hydrogel created 863,040 conductive pathways across the membrane. The engineers noted that they could extract 1.7 million pieces of data from this design when using all available pathways.

The next step was to register the data gathered during certain actions, such as tapping the finger, squeezing the hand, a deep cut, or a gentle touch. This data was then fed into a purpose-built machine learning algorithm.

The AI used this data model to determine what sensory input held precedence over other actions. For example, a deep cut is more important than a light prick of the finger. They also noted that the system supported customization of these preferences.

Depending on the robot’s task, engineers can adjust what senses have priority. For example, you may want the sensitivity turned up on an agricultural robot feeling for ripeness, but the temperature sensors turned down as it’s working in the summer sun. Additionally, you can calibrate the device using a human touch.

Testing the Robotic Skin’s Capabilities

The engineers conducted several physical tests to ensure the device performed as expected. These tests included running their fingers gently over the hand, applying heat to the device, and squeezing it at different levels to register the difference between light and harmful actions. They also tested if it could sense multiple different types of touch simultaneously. In one instance, the hand was cut with a medical scalpel.

Robotic Skin Test Results

The test demonstrated that machine learning successfully determined the different actions in real time. The hand could make distinctions between human touch versus non-animate objects.  It could also determine weather conditions based on a combination of factors such as heat and moisture, just like a human.

Key Benefits of Robotic Skin Technology

There are several benefits that the robotic skin brings to the table. For one, it allows robots to interact with the physical world in a human-like way. Robots in the future will be able to tell if they bump into you or something randomly and adjust.

These devices will be set up to understand how much pressure they are applying to items based on the item’s response, opening the door for more effective medical devices and much more. The highly sensitive nature of the robotic skin makes it ideal for a wide range of applications.

Low-Cost

One of the biggest benefits of the new robotic skin design is that it’s easy to fabricate. Traditional robotic skins relied on multiple sensors, precisely wired into place. This approach costs a lot more than utilizing conductive hydrogel and lacks the flexibility to mold into nearly any shape. This new approach will make touch-sensitive robotics available to more devices.

Robotic Skin is Reusable

Another key benefit of the robotic skin design is its ability to be reused. The conductive material can be melted and molded as seen fit. The hydrogel can fit complex shapes without the need to run internal wires. This reusable approach will help prevent future e-waste from robotics.

Durable

The robotic skin design is far more durable than its predecessor. The fact that there aren’t sensors located throughout the device limits the possibility of relevant damage. This structure provides support for flexible and reusable components that can hold up under heavy stress. They can also be repaired utilizing hydrogel without the need to repair complex internal sensors.

Customizable

The customizability of the robotic skin is a huge plus. The device can react differently to different types of stimuli depending on the job. For example, you could have a robot equipped to ignore thorns when picking roses but be hypersensitive to the moisture or feel of the stem.

Applications and Timeline for Robotic Skin

There are several applications for robotic skin. The obvious use would be to enhance the capabilities of humanoid robots. These devices already resemble humans in their structure and appearance. Adding a layer of touch could help to make them even more human-like and relatable.

Prosthetics

This study could help to make prosthetics that can feel become a reality. The use of soft, flexible hydrogel as skin will help to make these medical devices seem more real for the wearer and those around them. Additionally, they can be fitted to provide the same amount of feel as the lost body part or even enhanced with additional sensory input.

Automotive Sector

The automotive sector has been at the forefront of robotics integration. This study could help make robots that can work alongside humans safely. These devices could recognize when they get close to or interact with a human and react like a co-worker would if they were to bump into you in an accident and pull away.

Agriculture Sector

Touch-sensitive robots are already being used to determine key aspects of agriculture. The device could utilize robotic skin to reduce its weight, increasing its operational times and performance. These systems could also integrate purpose-built AI models to improve performance and ensure results.

Robotic Skin Timeline

It could be 5-10 years before robotic skin makes its way to the market. The engineers will complete more work on their design and seek out partnerships and funding. If successful, you could see robotic skin on devices in warehouses in the next 5-7 years. If all goes well, medical devices could utilize this option within the next decade.

Who Developed Robotic Skin?

University of Cambridge and University College London (UCL) engineers worked together to make this study a reality. The paper lists David Hardman, Thomas George Thuruthel, and Fumiya Iida as the primary authors of the work. Notably, the study received support from the Royal Society,  the Engineering and Physical Sciences Research Council, the Samsung Global Research Outreach Program, and UK Research and Innovation.

Robotic Skin Future

The team now looks to improve their device’s durability and sensitivity. These next steps will include further researching materials for the unit and new sensor types. The goal is to help innovate and promote single-layer skins in sensitive systems and drive the use of multimodal sensory options in the future.

Investing in Robotics

The robotics sector has several high-level competitors. These companies range in their offerings from robots designed to monitor infrastructure to healthcare devices and military hardware. All of these sectors have seen considerable growth over the last decade, with analysts predicting even more in the future. Here’s one company positioned to capitalize on future adoption.

UiPath

UiPath (PATH -4.15%) entered the market in 2005. At that time, it went by the name DeskOver. The company’s founders, Daniel Dines and Marius Tîrcă, wanted to create reliable robotic automation software for the budding industry. The company’s first decade saw slow but steady growth.

In 2015, the firm moved to UiPath and expanded operations. The firm opened a headquarters in NY, with new offices in London, Bangalore, Paris, Singapore, Washington D.C., and Tokyo. In 2021, UiPath hosted one of the largest US software IPOs, where it secured $35B to further its operations.

Today, UiPath plays a vital role in robotic automation. Their software helps fleets of robots to remain in sync, operating safely with thousands of human coworkers and other devices. As such, those seeking exposure to the robotics industry should do more research on UiPath shares.

Latest UiPath (PATH) Stock News and Developments

Final Thoughts: The Future of Robotic Skin

The robotic skin concept could enhance how humans and robots interact in the future. This device could be integrated into existing bots, providing them with an upgraded understanding of their surroundings and new capabilities. For these reasons and many more, you have to commend these engineers on their hard work.

Learn about other cool robotics developments here.

Studies Referenced:

^{1. David Hardman et al., Multimodal information structuring with single-layer soft skins and high-density electrical impedance tomography.Sci. Robot.10,eadq2303(2025).DOI:10.1126/scirobotics.adq2303}