Researchers at Washington State University have developed a customisable electronic skin capable of mapping pressure and temperature across flat and curved prosthetic surfaces.
The modular system combines digital scanning, 3D modelling, additive manufacturing, laser cutting and machine-learning-assisted data processing. Its developers say it achieves approximately ten times finer spatial resolution than existing commercial glove sensors.
The research, published in Cell Reports Physical Science, could help address one of the major limitations of upper-limb prostheses: a prosthetic hand may grasp an object, but the user generally cannot naturally feel its temperature, texture or the pressure being applied.
A report published by The Good Men Project describes the technology as a possible foundation for future “bionic skin” combining sensing with haptic stimulation.
However, the current system is primarily a sensing platform. Further development is needed before the information collected by its sensors can be converted into meaningful sensation for a prosthesis user.
Why sensory information matters
Visual feedback plays a major role when operating a conventional prosthetic hand. Users often need to watch the terminal device closely to judge whether it has made contact with an object and whether the grip is secure.
This can make everyday tasks slower and more cognitively demanding. Excessive force may damage a fragile object, while insufficient force may allow it to slip.
Temperature presents another challenge. Without an effective feedback mechanism, the user may not immediately recognise that a prosthetic hand has contacted a hot surface. Sensors could potentially detect the risk, but the system must then communicate that information in a form the user can understand quickly.
Future sensory systems could help users determine:
- Whether the prosthetic hand has contacted an object
- How firmly the object is being held
- Whether grip pressure is evenly distributed
- Whether a surface is hot or cold
- Whether an object is beginning to slip
- Differences in surface texture
- Characteristics that help distinguish one material from another
Reliable sensory feedback could also reduce the need for constant visual monitoring and potentially make prosthesis use feel more intuitive.
A scan-model-print workflow
Electronic skins are not entirely new, but fitting sensor arrays over complex three-dimensional shapes remains difficult.
Flat sensor sheets may not conform evenly to fingers, joints and curved sections of a prosthetic hand. Stretching or adapting a standard array to a custom shape may also reduce sensing accuracy, create gaps or produce mechanically vulnerable areas.
The Washington State University team developed what it calls a scan-model-print workflow.
The geometry of the prosthesis is first digitally scanned. Researchers then use the resulting model to position sensor modules according to the shape and curvature of the intended surface. The modules can subsequently be produced in forms that match the individual prosthesis.
According to the peer-reviewed research record, this geometry-aware approach enables high-resolution pressure and temperature mapping across free-form surfaces.
For prosthetists and technicians already familiar with digital scanning and computer-aided design, the concept is recognisable: capture the actual geometry first, design around it and manufacture a patient- or device-specific form.
In this case, however, the output is not a socket or cosmetic cover. It is a sensing layer designed to follow the prosthesis surface.
Modular sensors connect without adhesives
The sensor modules are constructed as thin layered units containing interlaced pressure and temperature matrices.
Self-aligning snap-fit connections allow the modules to join together in a manner compared with construction blocks. This approach removes the need to rely entirely on adhesives and permits sensor coverage to be configured for different shapes and regions.
Modularity could offer several practical advantages:
- Damaged sections may eventually be replaceable individually
- Different areas can receive different sensor densities
- Coverage can be adapted to the geometry of a particular hand
- Modules may be reconfigured as the design evolves
- Manufacturing could be simpler than producing one continuous custom skin
These potential advantages still require validation under real-world conditions. Snap-fit electrical connections on a prosthetic hand would need to withstand repeated impacts, vibration, finger movement, sweat, dust, moisture and daily cleaning.
Machine learning processes the sensor data
High-resolution sensing creates another problem: large arrays produce substantial quantities of data.
If every pressure and temperature point must be processed individually, the system may become slow or require excessive computing power. Any delay between contact and feedback would reduce its value during functional tasks.
The researchers used a multihead neural network to calibrate and denoise the data in real time. The system aims to reduce errors at individual sensing points while preserving the ability to recognise patterns across the wider surface.
In laboratory testing, the platform could support the identification of textures and material properties. The researchers reported pressure and temperature resolution around ten times finer than that of the commercial glove sensors used for comparison.
This does not necessarily mean that the e-skin is ten times more clinically useful. Spatial resolution is one performance measure, while practical benefit also depends on reliability, latency, durability, feedback design and how accurately the user interprets the information.
Detecting a stimulus is not the same as feeling it
The most important distinction is between sensing and sensory feedback.
The electronic skin can detect and map pressure and temperature. It does not yet, by itself, restore a natural sense of touch to an amputee.
To achieve functional feedback, the detected information must be translated into a signal delivered to the user. Possible approaches include:
- Vibration against the residual limb
- Mechanical pressure
- Electrical stimulation of the skin
- Thermal cues
- Stimulation of peripheral nerves
- Surgically implanted neural interfaces
The research team is developing an actuator intended to convert the sensor data into stimulation that could communicate with nearby nerves.
This will be a major translational challenge. The feedback must be fast, understandable and proportional without becoming painful, distracting or mentally exhausting. It must also be integrated safely with the socket, electrodes, prosthetic controller and power system.
Questions before clinical adoption
The published study establishes an engineering platform, not a market-ready prosthetic product.
Future work will need to evaluate:
- Long-term durability during everyday prosthesis use
- Resistance to sweat, water, dust and cleaning agents
- Performance over moving finger and wrist joints
- Sensor drift and recalibration requirements
- Power consumption and battery implications
- Comfort, bulk and cosmetic acceptability
- Compatibility with existing prosthetic hands
- Repair procedures and replacement costs
- Feedback methods that users can interpret
- Functional outcomes during daily activities
- Safety and medical-device regulatory requirements
Clinical studies involving people with upper-limb loss will be particularly important. Laboratory identification of texture and material is encouraging, but the decisive test is whether the complete system improves task performance, confidence, safety and prosthesis acceptance.
Relevance for India
India has a growing ecosystem in additive manufacturing, flexible electronics, artificial intelligence, rehabilitation engineering and prosthetic design. The scan-model-print approach creates opportunities for collaboration between P&O schools, engineering institutes, hospitals and technology companies.
The modular manufacturing model may also be relevant to Indian service conditions. A system that can be configured and repaired using replaceable modules could eventually be more practical than a proprietary electronic skin that must be returned overseas when one area fails.
Local development should focus on more than reproducing the sensor array. Indian research teams could contribute by testing performance under heat, humidity, dust and intensive daily use, as well as by developing affordable feedback interfaces and repair systems.
Prosthetists and prosthesis users should be involved early. The most technically advanced sensing layer will have limited impact if it increases weight, reduces grip options, interferes with glove replacement or cannot be maintained locally.
Cost will also require careful analysis. The researchers describe 3D printing and laser cutting as relatively simple production methods, but no complete clinical price has been established. Manufacturing cost must eventually include electronics, calibration, software, feedback hardware, fitting, training, maintenance and regulatory compliance.
A step towards personalised bionic skin
The study provides a promising response to a genuine prosthetic challenge. It combines custom geometry with high-density multimodal sensing without requiring every prosthesis to use the same flat sensor sheet.
Its most significant contribution may be architectural: a method for mapping modular sensors to the unique surface of a prosthesis while processing their output in real time.
The technology does not yet give an amputee a natural sense of touch. It does, however, create an adaptable sensory layer that could form one half of that system.
The next task is to connect what the prosthesis detects with what the user can meaningfully perceive.
- The Good Men Project: Electronic Skin for Prosthetics Senses Temperature and Pressure
- Washington State University: Researchers Develop Electronic Skin for Prosthetics
- Research Record: Geometry-Aware Multimodal Sensing System
- Cell Reports Physical Science: Original Research Paper