Researchers from the University of Cambridge and University College London (UCL) have developed a low-cost, single-material electronic skin (e-skin) that can simultaneously detect heat, pressure, and damage, offering a promising step toward more human-like tactile sensing in robotics.
Their findings, published in Science Robotics, present a simpler yet powerful alternative to the complex sensor arrays typically used today.
Study: Multimodal information structuring with single-layer soft skins and high-density electrical impedance tomography. Image Credit: aerogondo2/Shutterstock.com
Background
Traditional robotic tactile systems often rely on combining multiple sensors—one for pressure, another for temperature, another for detecting damage—embedded within soft materials. While this multi-component approach can deliver functional results, it tends to be fragile, difficult to fabricate, and prone to signal interference. Replicating the seamless and multifunctional sensing ability of human skin remains a major challenge.
In contrast, this new study takes a streamlined approach: using a single, flexible, conductive material that can handle all types of tactile input.
The Approach: One Material, Many Signals
At the heart of the innovation is a gelatin-based hydrogel. Chosen for its electrical conductivity, softness, and ease of processing, the hydrogel can be melted, shaped into intricate forms, and integrated into robotic systems, closely mimicking the physical characteristics of real skin.
What makes this material particularly effective is its internal microstructure. Inside the hydrogel are roughly 860,000 tiny conductive pathways that respond uniquely depending on the type of contact—be it heat, pressure, or physical damage. When touched or altered, these pathways change their electrical impedance, providing rich, measurable signals.
To turn the hydrogel into a functional sensor, the team molded it into the shape of a human hand and connected it to a robotic system. They embedded 32 electrodes at the wrist, enabling the capture of over 1.7 million electrical data points, allowing for high-resolution, real-time tactile feedback.
Teaching the Skin to Feel
To make sense of the complex signals generated by the hydrogel, the researchers used machine learning.
They ran the e-skin through a series of controlled tests, applying heat with a heat gun, pressing it with fingers and robotic arms, and even simulating injury by making cuts with a scalpel. These interactions produced distinct electrical patterns that the algorithms learned to identify and differentiate.
Over time, the system became adept at recognizing each type of touch with impressive accuracy, even when stimuli were combined or applied with varying intensity.
Why it Matters
The sensor's ability to handle multiple stimuli through a single material, and still deliver precise, real-time feedback, is a big leap forward. Unlike traditional multi-sensor setups, this design bypasses the problems of interference and fragility. It’s also more scalable: simpler to produce, more cost-effective, and easier to adapt for different robotic applications.
Machine learning plays a crucial role here. By interpreting subtle differences in the hydrogel's signals, the algorithms can extract useful information while ignoring background noise. This allows the sensor to operate quickly and reliably in real-world conditions.
Equally important is the hydrogel's durability. It can bend, stretch, and recover from minor damage without losing function, qualities that make it especially suitable for robotic hands or flexible surfaces that interact with unpredictable environments.
Looking Ahead
While there’s still room to refine sensitivity and improve long-term durability, this research marks a significant advance in tactile robotics. It shows that a single, well-designed material can serve multiple sensing roles, bringing robotic systems closer to the dexterity and sensitivity of human skin.
Journal Reference
Hardman D. et al. (2025). Multimodal information structuring with single-layer soft skins and high-density electrical impedance tomography. Science Robotics, 10,eadq2303. DOI: 10.1126/scirobotics.adq2303, https://www.science.org/doi/10.1126/scirobotics.adq2303