humanoidsdata.com

Search

Search datasets, articles, and glossary terms for humanoids and embodied AI.

← All glossary terms

Hardware & control

Tactile sensing

Tactile sensing is the detection and measurement of physical contact properties at a robot's surface or contact interface. Depending on the sensor, it can report pressure or force distribution, contact location, shear, vibration, slip, texture, temperature, or deformation. Tactile data complements vision by measuring interactions that may be hidden at the point of contact.

Also known as: robot tactile sensing, robotic tactile sensing

Updated

Contact produces information that cameras can miss

Dahiya and colleagues define tactile sensing around detecting and measuring contact parameters over a specified area, with sensor-level processing before higher-level interpretation. Robot implementations range from single contact switches to fingertip arrays, optical tactile sensors and large-area electronic skin.

Vision may show that fingers surround an object while hiding whether the object is slipping or how pressure is distributed. Tactile sensing can reveal contact onset, local force changes, vibration or deformation during grasping, insertion, wiping and other contact-rich tasks.

Tactile, force-torque and proprioceptive signals differ

A tactile array measures contact locally at or near a surface and may preserve spatial structure. A wrist force–torque sensor measures the resultant forces and moments transmitted through the wrist, but usually cannot identify the complete pressure pattern across several fingertips. Motor currents and joint torque estimates are internal signals that can imply contact without measuring the interface directly.

Proprioception describes the robot's own configuration and motion, such as joint position and velocity. Tactile sensing is generally exteroceptive because it reports interaction at the boundary between body and environment. A manipulation system can combine all of these streams.

Tactile data needs calibration and timing

Raw values are difficult to compare without sensor type, location, units, range, resolution, sampling rate, calibration and drift information. A dataset should also align tactile samples with camera frames, hand pose, actions and outcomes.

Contact sensors can saturate, wear, detach or respond differently after remounting. Training records should preserve those limitations and any filtering or normalisation. A tactile channel adds value when its physical meaning and relationship to the task remain recoverable.

Sources