KiriSense technical insights
Normal Force, Shear Force and Slip in Robotic Manipulation
Normal loading, tangential loading and sliding are related, but they are different physical quantities and contact behaviours.
By KiriSense Ltd · Published 2026-10-06 · Updated 6 October 2026. Technology lead: Kangsheng Bretherton-Liu.
Normal and shear force describe directions
Normal force acts approximately perpendicular to a contact surface. Shear force acts tangentially. On a flat fingertip interface, normal loading presses into the surface while shear loading acts along it. For curved or changing contacts, the local surface orientation matters when defining these directions.
Force is not the same as pressure. Pressure describes loading per unit area. A contact patch may have an uneven distribution even when a controller reports one total normal force. Likewise, a resultant shear force does not reveal every local motion within the patch.
Why robots need shear information
When a robot lifts an object, the contacts must support tangential loading as well as the forces squeezing the object. Turning or repositioning the object changes that loading. Shear information can help a manipulation system identify lateral loading and changes in contact behaviour.
Normal-only feedback can tell a controller that it is pressing, but may not describe the direction of the load along the surface. Multi-axis sensing can add that distinction. Its usefulness depends on calibration, coupling between axes, contact geometry and the controller’s task. See fingertip sensing for dexterous hands.
Shear loading does not mean that an object is slipping
A sticking contact can carry shear force without sliding. Slip means relative movement between contacting surfaces. Incipient slip is the onset of local sliding: parts of a contact patch can move while other parts remain stuck, before gross slip develops.
Friction links loading to whether a contact can stick, but the relevant behaviour depends on the materials, surface condition and force distribution. A simple friction model can be useful for reasoning about a grasp; it does not replace measurements of the actual interface. Detecting nonzero shear alone is not a validated slip detector.
How robots detect slip
Tactile systems can look for contact motion, changing deformation patterns, vibrations or other signals associated with sliding. Image-based sensors can track changes across a contact surface. Other sensors use changes in force or deformation over time. The chosen method must be checked against actual relative motion and relevant objects.
Dong and colleagues demonstrated monitoring incipient slip using GelSight measurements in a grasp-control setting. Their work provides a research example of detecting local contact changes and acting before gross slip. It should not be read as proof that any sensor with a shear response can perform the same task.
From a signal to a useful manipulation system
A complete system must acquire a signal, interpret contact state and update the robot’s action. Time alignment matters when comparing force with movement. Increasing grip force may help some slipping grasps, but can damage a delicate object; an appropriate response may instead involve repositioning or stopping.
Validation therefore needs representative loads, contact locations, materials and motions. Force accuracy, cross-axis coupling, hysteresis, repeatability and slip-detection error are different measurements. They cannot be inferred from visually distinct optical signatures alone.
What geometry-encoded sensing contributes
KiriSense uses physical geometry to convert different loading conditions into distinguishable optical responses before downstream interpretation. Its laboratory proof of concept distinguishes normal and directional shear loading qualitatively. Those observed signatures establish the sensing principle. Quantitative force calibration and slip-detection validation follow.
Quantitative characterisation is the next step towards relating the signatures to applied loads. Testing with moving contacts and a robotic controller would then establish which slip or contact-state information can be recovered. This preserves the distinction between the demonstrated normal/shear response and future robotic feedback capabilities.
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 et al. (2018), Maintaining Grasps within Slipping Bound by Monitoring Incipient Slip
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.