KiriSense technical insights
What Is Robotic Tactile Sensing? Vision, Touch and Contact
Robotic tactile sensing gives a robot information about physical contact that cannot always be obtained from vision alone.
By KiriSense Ltd · Published 2026-06-02 · Updated 6 October 2026. Technology lead: Kangsheng Bretherton-Liu.
What is a tactile sensor?
A tactile sensor converts a physical interaction at a contact surface into a measurable signal. In robotics, that signal can describe loading, deformation, contact location or motion, depending on the sensor architecture. The controller uses the signal to decide whether and how to change the robot’s action.
A robot fingertip sensor places this measurement near an object-hand contact. Sensors can also cover gripper pads or other surfaces. A single force value, a spatial pressure map and a detailed optical contact image are different kinds of information; none is a complete description of every contact.
What vision and touch each contribute
Vision supports object recognition, scene understanding and planning a grasp. It can observe motion outside a contact. Once fingers close, the important interface may be occluded. Touch samples what happens locally when a robot makes contact, including loading changes that may not be visible to an external camera.
Touch therefore complements vision. It does not replace the wider view of the scene. A robot can use vision to approach an object and tactile feedback to adjust its grasp after contact. Inserting a component or turning an object inside a hand needs both geometric context and feedback about the interaction.
Normal force, shear and contact state
Normal force acts approximately perpendicular to a contact surface. Shear force acts tangentially. A fingertip pressing into an object creates normal loading; lifting or sliding can also create shear loading. Pressure is force distributed over area, so it should not be treated as interchangeable with a total force measurement.
Contact state includes whether contact exists, whether it is sticking or sliding, and where or how it changes. Shear loading can exist while a contact sticks. Detecting slip requires evidence of movement or a validated signal pattern, rather than simply detecting a nonzero shear force. See normal force, shear force and slip in robotic manipulation.
How robotic fingertips detect contact
Resistive, capacitive and magnetic sensing elements can convert deformation or loading into electrical measurements. Arrays place multiple elements across a surface. Some designs sense more than one force direction, including shear. Readout, calibration and mechanical construction determine what can be recovered.
Optical tactile sensors use light to observe a mechanical response. Image-based systems such as GelSight and DIGIT observe a deformable surface with a camera. KiriSense instead develops a geometry-encoded mechanical and optical response, with compact tactile signals as its design proposition. Compare optical tactile sensing approaches without assuming that one architecture is best for every task.
Useful feedback requires a control task
Dexterous manipulation involves changing an object’s pose or grasp through coordinated contact. More sensing data is useful only when it helps that task. A gripper may need to adjust grip as an object is lifted; a dexterous hand may need to coordinate several contacts during in-hand movement.
For embodied AI, tactile measurements can supply evidence about the consequences of physical actions. Integrating them requires time alignment, calibration, suitable training data and a controller that responds appropriately. Robot hands and humanoids illustrate why fingertip feedback matters and why sensing alone is not dexterity.
KiriSense’s approach to robotic touch
KiriSense Ltd develops geometry-encoded robotic tactile sensing for robot fingertips and dexterous manipulation. Physical geometry encodes tactile information before downstream computation. The architecture is designed to sense normal, shear and slip information through compact signals rather than a conventional high-bandwidth tactile image stream, targeting low-cost, low-data and low-latency feedback.
Explore KiriSense’s geometry-encoded tactile sensing architecture. Read the latest proof-of-concept results for the dated evidence from the company’s research collaboration with the University of Sheffield. Technology origin and company-owned IP and technology lead Kangsheng Bretherton-Liu explain who is behind it.
Primary research
Dong, Yuan and Adelson (2017), Improved GelSight Tactile Sensor for Measuring Geometry and Slip
The references support the general science and examples discussed above. KiriSense-specific claims are described in its linked proof-of-concept report.