As robots move from structured automation into environments where they need to grasp, manipulate and interact with unpredictable objects, vision and proprioception alone leave a significant sensing gap: what is actually happening at the point of contact?
XELA Robotics approaches this problem with uSkin, a thin, compliant tactile sensing platform designed to give robots distributed 3-axis touch.
Rather than measuring only normal pressure, each sensing point measures X, Y and Z displacement, providing information about both normal and shear forces. This means a robot can detect not only how hard it is gripping something, but also how the object is moving within the grasp.
The architecture is particularly interesting from an engineering perspective. uSkin combines a soft elastomer layer with an array of compact sensing points, allowing the sensor to conform to the object being grasped while remaining resilient to overload. Individual sensing points are approximately 4×4 mm, with sensor packages starting at around 4mm thickness. XELA also offers curved and custom geometries, allowing sensing to extend beyond fingertips into phalanges, palms and other robotic surfaces.
Most tactile technologies involve a trade-off between sensitivity, spatial resolution, mechanical compliance, robustness and integration complexity. Resistive and piezoresistive sensors can be relatively straightforward and scalable, while capacitive approaches can provide good sensitivity but introduce challenges around noise, environmental effects and electronics. Optical tactile systems can deliver very high spatial information, but typically require cameras, illumination and additional mechanical volume.
XELA’s approach is distinctive in combining several characteristics in one package:
One particularly notable detail is the calibration architecture. XELA can individually calibrate sensing points to compensate for variation between them, with the aim of improving force accuracy and reproducibility. The system also incorporates temperature sensing and compensation for magnetic interference.
The interesting engineering opportunity isn’t simply adding another sensor to a robot. It is creating a closed-loop interaction layer between the robot and the physical world.
A tactile system that can provide spatially distributed normal and shear information gives the control system access to signals that vision cannot directly observe such as contact force, contact distribution, object movement within a grasp and the onset of slip.
For dexterous manipulation, that distinction becomes important, opening up a wide range of development possibilities.