UltraSense proposes ultrasound for scalable, durable robotic tactile sensing

Physical AI is moving from digital reasoning into real-world interaction. Robots can see, plan, navigate, and move with sophistication. Yet manipulation depends on reliable touch, creating an opportunity for ultrasound.

The question is no longer whether robots need tactile sensing. For humanoid hands, dexterous grippers, logistics robots, service robots, and industrial systems, touch must detect contact, map force, sense shear and slip, and understand material interaction. The question is which tactile architecture can scale.

Many tactile and electronic-skin approaches rely on flexible electrical sensing layers placed on or near the contact surface. These include capacitive, piezoresistive, piezoelectric, triboelectric, and impedance-based structures integrated into elastomers, flexible substrates, conductive networks, or multilayer films. These technologies have advanced the field.

However, for high-duty-cycle robotic fingers and grippers, electronic skin faces a scaling challenge: the sensing layer is often located close to the harshest mechanical environment on the robot.

That environment includes repeated compression, abrasion, contamination, humidity, temperature variation, cleaning exposure, and material aging. Even when encapsulated, a surface-coupled stack remains tied to the outer skin. Over time, that can introduce wear, hysteresis, creep, delamination, baseline drift, and recalibration burden.

For a laboratory prototype, these issues may be manageable. For a commercial robotic hand operating over millions of contact cycles, they become central to product viability. UltraSense believes the better path is protected sub-surface ultrasound.

Editor’s note: Physical AI is among the session topic tracks at RoboBusiness 2026, which will be on Oct. 20 and 21 in Santa Clara, Calif. Register now to attend.

The scaling problem with surface-coupled electronic skin

A robotic fingertip repeatedly presses, slides, rubs, and impacts the world. In a humanoid hand or dexterous gripper, it must provide compliance and friction while surviving long use. This creates a tradeoff for exposed or near-surface electronic skin.

If the sensing layer is close to the outer surface, it can be affected by mechanical wear. If the layer is padded for protection, spatial resolution and sensitivity can degrade. If the elastomer ages, the signal can drift. If the stack experiences repeated shear, adhesion and interlayer stability can become concerns.

The issue is not visible wear; it is lifetime signal integrity. A tactile system must remain accurate, repeatable, and calibrated over time. In robotics, the sensor that works on Day 1 must still provide useful contact data after months or years.

Recent advances in thin-film force sensing show that the industry is moving beyond simple single-effect sensors. Force sensing resistors can drift as conductive networks change under repeated compression, while capacitive sensors can be affected by parasitics, environment, and mechanical constraints.

Newer impedance-based approaches capture broader electrical response by measuring combined resistive and capacitive behavior. This reinforces an important distinction: Reliable robotic touch needs richer information than binary contact or a single force value.

However, these approaches still generally depend on a functional sensing layer in the mechanical load path. For robotic hands and grippers, that layer remains part of the wear, compression, shear, and contamination environment.

The outer surface should be optimized for friction, compliance, sealing, abrasion resistance, and durability while the sensing element remains protected. That is the advantage of ultrasound.

Ultrasound moves sensing below the damage zone

Ultrasound enables a different tactile architecture. Instead of relying on surface electrical deformation, an ultrasound sensor can interrogate the material stack from below. Acoustic waves propagate through the structure, reflect from internal boundaries, and change in response to contact-induced deformation.

When an object presses into a compliant surface, the geometry of the stack changes. Acoustic path length changes. Echo timing changes. Reflection amplitude and frequency content can change.

Under shear, internal displacement and boundary conditions can shift laterally. These acoustic changes can infer what is happening at the contact interface.

Ultrasound extends multi-parameter sensing into the acoustic domain. Instead of measuring only a local electrical change inside a deformable film, ultrasound analyzes the broader acoustic response of the material stack.

Time-of-flight (ToF), reflection amplitude, attenuation, frequency response, phase, and acoustic impedance mismatch provide information about contact, deformation, force distribution, shear, and material interaction. This makes ultrasound a multi-parameter, not single-effect, sensor architecture.

This allows sensing below the outer skin, away from direct abrasion and surface wear. The robot designer can choose an outer material — elastomer, polymer, glass, metal, leather, fabric, or another engineered surface — while ultrasound reads contact through the stack.

UltraSense’s platform follows this principle: Let the outer surface handle mechanics, while ultrasound delivers tactile intelligence below the surface.

Better performance: touch, force, shear, slip, and material detection

Robotic touch is not binary. A useful tactile system must provide multiple information layers.

First, it must detect touch and contact location. Ultrasound can identify contact by observing how the acoustic response changes when the surface is mechanically perturbed.

Second, it must provide force mapping. A single force value is not enough. A robot needs to know whether force is centered or off-axis, whether the contact patch is broad or concentrated, and whether the load is shifting. Ultrasound can infer deformation through ToF and echo changes, enabling localized force mapping.

UltraSense has demonstrated ultrasound tactile sensing with 500 µm spatial resolution through an elastomer layer. At that resolution, the system can resolve fine structures, including ridge and valley patterns in a contact profile. Our company has also demonstrated localized compressive force profiling with approximately 1.25 mN precision.

Third, the system must understand shear and slip. For many manipulation tasks, shear matters more than normal force. A robot may press hard enough to hold an object, but if tangential slip begins, the grasp can fail before vision detects motion. Humans use shear and micro-slip cues constantly; robots need similar feedback.

UltraSense’s subsurface ultrasound approach can track lateral displacement within the material stack to infer shear-related behavior. In demonstrations, our technology has shown shear-force inference with an approximately 5 mN noise floor.

Fourth, ultrasound can support material/contact classification through acoustic impedance. Every material presents a different acoustic impedance based on density and sound velocity. When ultrasound reaches the contact interface, reflected and transmitted energy changes depending on whether the object is metal, glass, plastic, rubber, fabric, skin, foam, or another soft material.

By analyzing reflection amplitude, echo signature, attenuation, frequency response, and time-domain changes, the system can infer both tactile response and the acoustic nature of the surface being touched.

Ultrasound offers cost-effective scaling through platform reuse

To scale, tactile sensing cannot remain a custom research assembly. It must become a manufacturable platform with repeatable electronics, calibration, firmware, packaging, and integration.

UltraSense brings a foundation built through automotive human-machine interaction (HMI) and AI device experience. Automotive applications require reliability, temperature performance, vibration tolerance, moisture resistance, manufacturability, and repeatability. AI devices require small form factors, low power, premium materials, sealed construction, and intuitive interaction.

The company has shipped more than 4 million units into automotive applications, providing production experience in ultrasound sensing, mixed-signal ICs, firmware, calibration, test, and integration. The same platform principles now extend into tactile intelligence for physical AI.

This platform reuse matters. A scalable tactile solution cannot depend on complex lab-grade assemblies or fragile sensor films requiring frequent replacement. It needs compact modules, integrated electronics, repeatable manufacturing, and edge processing.

UltraSense’s ASIC roadmap reflects this experience. Without disclosing implementation details, the platform is designed as an intelligent SoC-based architecture with embedded processing for tactile zones.

The goal is to create a local tactile intelligence layer that can detect touch, estimate force maps, infer shear and slip, classify contact state, detect material response, and provide actionable information to the robot controller.

Material flexibility and IP protection

In HMI, UltraSense has enabled touch on real product surfaces, including metal. Conventional capacitive touch struggles with solid metal interfaces, where electric fields are shielded or distorted. Ultrasound changes the interaction model by sensing through acoustic propagation rather than surface electric-field modulation.

The same applies to robotics. A robotic fingertip may need an elastomer surface for compliance and friction. An industrial gripper may require a rugged polymer. A medical or service robot may need a cleanable surface. AI devices may use metal, glass, leather, or fabric for premium industrial design. A tactile platform should adapt to these surfaces rather than force designers into one fragile sensing material.

As tactile sensing moves to commercial deployment, IP becomes critical. UltraSense has built a strong issued and pending patent portfolio around ultrasound-based touch, force sensing, ToF tactile sensing, calibration, and related system architectures, including tactile sensing under a cover layer, ultrasonic touch systems, force-measuring integrated circuits, and calibration and test methods.

Ultrasound could enable the next layer of physical AI

The next generation of physical AI will need more than better vision and motion. It will need reliable contact intelligence.

Robots must detect forces, classify contact, and adapt in real time. They must do this through real materials, under real environmental conditions, over long operating lifetimes, and at manufacturable cost.

UltraSense’s platform is designed to meet this challenge. By combining protected sub-surface sensing, high-resolution force mapping, shear and slip inference, acoustic impedance-based material response detection, custom ASICs, edge processing, and a strong IP portfolio, UltraSense is positioning ultrasound as a foundational tactile intelligence platform for physical AI.

About the authors

Mo Maghsoudnia is CEO of UltraSense Systems, and Hao-Yen Tang, Ph.D., is chief technology officer at the San Jose, Calif.-based company. Founded in 2018, UltraSense said it is building a user interface that differentiates products and enables new applications using patented 3D ultrasound technology.