Highly-Sensitive Electronic Skin Allows Robots to Feel Heat, Pain and Pressure

Typically, different types of sensors are needed to capture various forms of touch: one kind to detect pressure, another to measure temperature, and so forth. These sensors are then integrated into flexible, soft materials. However, these signals often interfere with one another, and the materials themselves are prone to damage. Now, scientists have developed a cost-effective, robust, and highly responsive robotic skin that fits over robotic hands like a glove, allowing robots to perceive their environment in a way that mimics human sensation. Unlike conventional robotic touch systems that rely on sensors embedded in small regions and require multiple sensors for different stimuli, the new electronic skin operates as a single, unified sensor across its entire surface, making it more similar to natural human skin.

The flexible and conductive skin, developed by researchers from the University of Cambridge (Cambridge, UK) and University College London (UCL, London, UK), is simple to manufacture and can be melted and reshaped into a variety of complex forms. The material is capable of detecting and processing multiple types of physical input, enabling robots to engage more naturally and effectively with their environments. While not yet matching the sensitivity of human skin, the new robotic skin can sense input from over 860,000 minuscule pathways within the material. This allows it to differentiate between various kinds of contact—such as a finger tap, exposure to heat or cold, physical damage like cuts or punctures, or multiple simultaneous touch points—using only one material.

To make the robotic skin more efficient in recognizing different types of touch, the researchers employed a combination of physical testing and machine learning algorithms. These techniques helped the material learn which of its internal pathways were most relevant to specific kinds of contact. The design relies on a single type of sensor capable of responding differently to varied stimuli, a method known as multi-modal sensing. Although it is more complex to pinpoint the source of each signal, multi-modal materials are generally easier to produce and more durable. The team used a soft, elastic, and electrically conductive hydrogel made from gelatin, which they melted and molded into the shape of a human hand. They experimented with different electrode placements to identify which configurations yielded the most informative data. Using just 32 electrodes placed at the wrist, they were able to collect over 1.7 million data points from across the hand, thanks to the dense network of conductive pathways.

To evaluate the skin’s effectiveness, the team conducted a range of touch tests. They exposed the skin to a heat gun, pressed it using both human fingers and a robotic arm, applied gentle touch, and even sliced it with a scalpel. The information collected during these tests was then used to train a machine learning model capable of interpreting the different forms of contact. Their findings are detailed in the journal Science Robotics. While the technology has clear potential for use in prosthetics and humanoid robots that benefit from tactile sensitivity, the researchers also suggest broader industrial applications. They plan to enhance the material’s durability and continue testing it in real-world robotic scenarios.

"We're not quite at the level where the robotic skin is as good as human skin, but we think it's better than anything else out there at the moment," said co-author Dr. Thomas George Thuruthel from UCL. "Our method is flexible and easier to build than traditional sensors, and we're able to calibrate it using human touch for a range of tasks."

Related Links:
University of Cambridge
UCL