top of page
Immagine del redattoreAndrea Viliotti

Digitizing Touch: The New Frontier of Digital Touch

The technology of artificial touch has made significant advancements thanks to Digit 360, a platform developed by Meta in collaboration with researchers from prestigious academic institutions, including FAIR at Meta, the LASR Lab at TU Dresden, CeTI at TU Dresden, and the University of California, Berkeley, with contributions from Mike Lambeta and Roberto Calandra. This device represents a step forward in replicating the complexity of the human sense of touch through multimodal implementation and the use of integrated artificial intelligence. The innovation is not only about sensory perception but also about the ability to process data in real-time, opening new scenarios for robotics, telepresence, and virtual reality.

Digitizing Touch: The New Frontier of Artificial Touch
Digitizing Touch: The New Frontier of Artificial Touch

Digit 360: A New Definition of Artificial Touch

The sense of touch is fundamental for interaction with the environment and is one of the most complex senses to replicate artificially. The ability to perceive the shape, texture, weight, density, and elasticity of objects through touch is crucial for both exploring the surrounding world and performing everyday actions. Human tactile perception is based on a series of specialized receptors, such as mechanoreceptors, thermoreceptors, and nociceptors, which provide a detailed understanding of physical stimuli. Each type of receptor is specialized in capturing a particular type of information: pressure, temperature, pain, and vibration.


Until now, artificial tactile sensors had a limited ability to detect complex and multimodal signals, failing to achieve the perceptual richness of human touch. Many previous systems were optimized to detect single contact modes, such as pressure or vibration, but lacked a holistic perspective that simultaneously accounted for the different perceptual components. This lack of multimodal integration has hindered progress toward a true digitization of touch equivalent to the human experience.


Digit 360 represents a significant change in this landscape as it is designed to integrate different sensory modalities: pressure, vibration, heat, and even the detection of volatile chemical compounds. This multimodality enables a complete and detailed perception of contact, allowing an understanding of an object not only through its shape and the force needed to manipulate it but also through subtler signals such as temperature variations and odors that the object may emit. For example, the ability to detect a change in temperature when touching a material can provide information on its thermal conductivity or state (e.g., if an object is hot or cold to the touch).


Another crucial aspect of human touch that Digit 360 attempts to replicate is the ability to perceive shear forces, which are fundamental for understanding how an object moves when it is grasped or manipulated. These shear forces allow humans to modulate grip force based on the perceived risk of an object slipping. Digit 360 can perceive and measure shear forces with a precision of 1.27 mN, enabling robots and devices to adjust their grip in real-time to avoid dropping or damaging fragile objects. This level of precision is crucial for applications where contact must be accurately controlled, such as in surgical operations or the manipulation of small electronic components.


A distinctive element of Digit 360 is its ability to process sensory data through a neural accelerator integrated directly into the device. This allows it to emulate a sort of "artificial reflex," similar to the human reflex arc, in which the response to a specific stimulus occurs locally without having to be processed by the system's "brain," i.e., the main control center. This architecture allows for a rapid response and a reduction in latency times, which is essential for tasks where even a brief delay could compromise the outcome of the interaction. Specifically, the integration of Edge AI reduces overall latency from 6 ms to just 1.2 ms, improving the device's efficiency and responsiveness.


Moreover, the ability to perceive vibrations up to frequencies of 10 kHz allows Digit 360 to distinguish materials with different surfaces simply by touching or rubbing them. This feature is particularly useful for identifying the texture of a material, determining whether it is rough, smooth, soft, or hard. Vibration perception also helps identify changes in contact, such as incipient slipping, and distinguish the different surface properties of manipulated objects.


Another feature that makes Digit 360 innovative is its ability to detect volatile chemical compounds through sensors that operate like an "artificial nose." This function can be used, for example, to identify the presence of potentially hazardous substances or determine the quality of food. Odor detection, combined with other sensory modalities, further expands the contextual understanding of the manipulated object, providing information that goes beyond what can be obtained through touch or sight alone. The device can detect odors with an accuracy of 91%, offering a precise classification capability of objects based on their chemical characteristics.


These advancements are not ends in themselves but serve to bridge the gap between the physical and digital worlds, making possible the interaction between machines and physical objects with a precision and sensitivity never seen before. Digit 360 aims to replicate the level of human perception in contexts where robots must operate autonomously, realistically, and safely in unstructured and complex environments. This development allows robots to interact with their environment in a way more similar to humans, enhancing their manipulation capabilities and adaptability.


Technical Features of Digit 360

The key element of Digit 360 is an artificial fingertip made of advanced materials, sensitive to both temporal and spatial stimuli. The fingertip is equipped with high-resolution sensors, with approximately 8.3 million "taxels" (tactile pixels) that provide very high spatial resolution. These sensors can perceive minimal details up to 7 micrometers, allowing for the distinction between surfaces with a very slight difference in roughness or structure. Accuracy in force detection has been measured with median errors of 1.01 mN for normal forces and 1.27 mN for shear forces, with variations depending on the position within the sensor.

The characterization of the material from which the sensor is made was performed using dynamic mechanical and thermal analysis (DMTA), determining an average Young's modulus value of 2.6 ± 0.74 MPa. This detail is crucial for understanding the strength and flexibility of the material that makes up the artificial fingertip, as well as ensuring an appropriate response to different tactile stresses.


Regarding forces, Digit 360 can measure normal forces with a precision of 1.01 mN and shear forces with a precision of 1.27 mN. The device is also capable of perceiving vibrations up to 10 kHz, allowing for the identification of different material properties during rubbing or dynamic interaction. This capability is made possible through the integration of MEMS microphones and pressure sensors that capture high-frequency signals, thus enriching the device's multimodal perception.


From an optical perspective, the system uses a hyper-wide-angle lens developed specifically to capture omnidirectional tactile images. This allows for a 360-degree view of the tactile interaction, with illumination provided by eight individually controllable RGB LEDs. The lighting system is designed to ensure uniform light throughout the volume of the sensor gel, minimizing artifacts and maximizing the contrast between deformations. This type of configuration optimizes the quality of the images acquired, reducing the need for software corrections and improving sensitivity in detecting deformations.


The sensor surface is made of an elastomer coated with a thin reflective layer, which helps improve optical resolution and capture deformations more accurately. The coating is achieved through a chemical deposition process, allowing for the creation of an extremely thin layer of reflective material, improving the sensitivity and resolution of the fingertip. This layer is about 6 micrometers thick, allowing even small deformations to be captured and increasing the sensor's ability to perceive details at a microscopic level.


The Digit 360 platform also includes a thermal detection system and volatile chemical compound detection. The thermal sensor allows for measuring heat variations during contact, useful for understanding the object's state, while the detection of chemical compounds helps determine the presence of specific elements, such as humidity or chemical residues. This type of sensor provides information that can be used for safety applications, quality control, or even for more realistic social interactions in telepresence contexts.


The multimodal detection capability of Digit 360, which includes visual, acoustic, pressure, thermal, and chemical signals, enables a much deeper understanding of manipulated objects compared to traditional tactile sensors.


Lighting and Optics

To improve spatial resolution and perception capacity, Digit 360 uses a customized optical system, including a hyper-wide-angle lens developed to capture omnidirectional tactile images. The lens is of the solid immersion hyper-wide-angle type, specifically designed for capturing tactile images and addressing the unique conditions of reflection and refraction within the elastomeric material. The lens has a diameter of 20 mm with an angular resolution of 1.1 micrometers.


The lighting system consists of eight individually controllable RGB LEDs distributed along the edge of the sensory area in a circular arrangement. Each LED can emit light with variable intensity and wavelength to adapt to the different needs of the interaction. The lighting configuration is designed to ensure uniform light distribution within the sensor volume, with optimal contrast between the elastomer's deformations and the background.


The design of the lighting system considered two main metrics: background uniformity and image-background contrast. The goal was to minimize reflection artifacts, improving the quality of the captured image. For this, techniques such as controlled light scattering on the reflective surface were adopted, with scattering parameters ranging from a completely smooth surface to a more diffused surface. This approach optimized touch sensitivity and ensured high image quality without areas of saturation or loss of detail in contact areas.


Moreover, the optical system integrates a reduction of chromatic aberrations by optimizing multiple focal lengths, thus improving the sensor's ability to detect deformations even under variable lighting conditions. Images are captured using a CMOS sensor with 1.1-micrometer pixels, ensuring high spatial resolution and an acquisition frequency of up to 240 Hz, essential for capturing dynamic variations in tactile contact.


Future Applications

The potential applications of the digital touch of Digit 360 are numerous and span various sectors, including robotics, medicine, virtual reality, telepresence, and advanced prosthetic systems. Below are some of the main potential applications and their implications.


Advanced Robotics

Digit 360 can significantly improve the manipulation capability of robots, particularly for delicate operations. Thanks to its high tactile resolution of 8.3 million taxels, robots equipped with Digit 360 can perform complex tasks requiring fine dexterity, such as manipulating delicate electronic components or picking fruit. The precision in measuring normal forces (up to 1.01 mN) and shear forces (up to 1.27 mN) allows for dynamic grip adjustment, minimizing the risk of damaging manipulated objects. For example, in agricultural automation, Digit 360 enables the picking of fruits without causing damage by identifying the exact pressure needed to handle each type of fruit.


Robotic Surgery

The high sensory resolution and the ability to process data in real-time thanks to the integrated neural accelerator make Digit 360 an ideal tool for robotic surgery. The rapid response (with latency times reduced to 1.2 milliseconds) allows surgical robots to adapt immediately to changes during operations, improving precision and patient safety. This is particularly useful in delicate operations, such as neurosurgery, where even a minor error can have critical consequences.


Virtual Reality and Telepresence

The ability to provide a multimodal perception of contact enables an unprecedented immersive experience in virtual reality (VR) and telepresence applications. Digit 360 can return tactile sensations that faithfully replicate real ones, such as the texture of surfaces or temperature variations upon contact. This opens up new opportunities for VR experiences, where users can perceive the virtual environment with an extremely realistic level of detail. In telepresence applications, such as telemedicine or remote exploration, Digit 360 allows operators to interact with the distant environment as if they were physically present, greatly improving the quality of interaction.


Bionic Prosthetics

In prosthetic systems, the implementation of Digit 360 offers patients the ability to perceive tactile sensations very similar to human ones. This can improve the quality of life for individuals with amputated limbs, restoring the ability to perceive texture, temperature, and pressure through prostheses. The sensory precision of Digit 360 allows patients to modulate grip force based on the delicacy of the object, reducing the risk of breaking or damaging fragile objects.


Safety Systems and Quality Control

The ability to detect volatile chemical compounds and temperature variations makes Digit 360 a useful tool in industrial safety and quality control contexts. For example, in industrial environments, the "artificial nose" of Digit 360 can detect the presence of potentially hazardous substances, such as harmful gases or chemical leaks, helping to ensure a safe working environment. In quality control, tactile sensitivity and the ability to detect minimal variations in materials make it possible to identify defects in products, improving the reliability of the production process.


Automation Industry

Digit 360 can be used to improve automation systems in production lines, where the ability to manipulate objects with precision is crucial. The integrated AI processing allows for rapid analysis and real-time response, enabling robots to make instant adjustments in assembly and handling activities. The ability to distinguish between materials of different textures and temperatures allows robots to select and assemble components more precisely and avoid costly errors.


Edge AI and Energy Efficiency

The integration of artificial intelligence algorithms directly into the device, without the need to transfer data to a central control system, improves operational efficiency. This Edge AI approach reduces latency and allows for an immediate response to stimuli. In distributed applications, such as managing a fleet of autonomous robots, this capability reduces overall energy consumption and improves system scalability. For example, robots operating in a warehouse can coordinate more efficiently, reducing downtime and increasing productivity.


Conclusions

The technology of digital touch represents more than just a technical advancement: it is a radical change in the relationship between humans and machines, posing profound questions about the fusion of biological perception and artificial sensory capabilities. With tools like Digit 360, we are approaching not only advanced automation but also a dimension of "sensory intelligence" in machines, a shift that could alter current business and industrial paradigms. It is no longer just a matter of replacing human actions but of rethinking how these technologies can modify the very value of certain human skills in production and decision-making processes, because, with Digit 360, machines acquire contextual sensitivity—the ability to "feel" and respond to context in real-time and with precision.


The crucial question for companies will be: what role should human skills have in a world where machines perceive like us, and perhaps even better than us? The precision of digital touch and the possibility of high-sensitivity actions in unpredictable contexts mean that many sectors, from manufacturing to logistics to medicine, may no longer need to rely on human sensitivity to ensure quality and safety. The impact will reflect on the skills required of people: we will need to shift human abilities to higher planes, such as strategy and supervision, while dexterity and perception may increasingly be delegated to technology.

In this context, tactile automation brings to the forefront the concept of empathic machines. The term may seem contradictory, but Digit 360 demonstrates that machines can be endowed with fine sensitivity to external stimuli and react like a "reflex," with zero latency, exactly like a human being. This is profoundly disruptive for businesses because it suggests that the effectiveness of technology will no longer be measured only in terms of speed or volume but also in terms of empathic response and adaptability, qualities that until now belonged only to the human world.


From a strategic perspective, the advancement in digital touch requires business leaders to rethink the ethics of automation. In which contexts is it legitimate to delegate tactile perception to machines, and in which, instead, does human presence and judgment remain indispensable? In contexts such as medicine or social care, the ability to "feel" an environment entails a responsibility that goes beyond mere operability, and companies will need to define new standards of responsibility and transparency in the use of digital touch.

The introduction of multimodal artificial touch should not, therefore, be seen merely as a functional evolution but as a redefinition of human interaction with technology and the meaning of contextual intelligence—a step that opens up new challenges and reflections on ethics and the value of sensitivity in production processes.


6 visualizzazioni0 commenti

Post recenti

Mostra tutti

Comments

Rated 0 out of 5 stars.
No ratings yet

Add a rating
bottom of page