UltraSense pitches ultrasound as robot touch fix; 500 µm resolution

UltraSense is making its case as physical AI moves from digital reasoning into real-world interaction, where manipulation depends on reliable touch. The company argues the industry has largely converged on flexible electrical sensing layers — capacitive, piezoresistive, piezoelectric, triboelectric, and impedance-based structures — placed on or near the contact surface. Those approaches have advanced the field, but for high-duty-cycle robotic fingers and grippers the sensing layer sits in the harshest mechanical environment on the robot.

That environment includes repeated compression, abrasion, contamination, humidity, temperature variation, cleaning exposure, and material aging, which can introduce wear, hysteresis, and recalibration burden. UltraSense's alternative is to interrogate the material stack from below with acoustic waves, letting the outer surface handle friction, compliance, and sealing. The company says this approach can infer touch, force mapping, shear, slip, and even material type through acoustic impedance, and it points to its automotive HMI background and 4 million shipped units as evidence it can manufacture at scale.

Whether ultrasound becomes a foundational tactile layer for physical AI will likely hinge on whether commercial robotic hands can operate over millions of contact cycles without the drift and recalibration that plague surface-coupled skins. UltraSense's claims are its own, and the broader test is whether robot makers adopt protected sub-surface sensing as a standard architecture rather than a laboratory demonstration.