Haptics: how touch works and how machines sense and recreate it

Touch is more than one sense

What we call touch is a group of senses working together. The skin reports contact, pressure, vibration, stretch, temperature and pain. Receptors in muscles, tendons and joints report the position of the fingers and the force they exert. Together these signals are called haptic information, from the Greek word for touching.

Without them, grasping would be clumsy. People whose fingertips have been numbed with an anaesthetic in experiments tend to grip much harder than necessary and still let objects slip. Touch is not an extra for the hand; it is part of the control loop.

The receptors in the skin

The glabrous skin of the palm and fingertips contains four main types of mechanoreceptors. In simplified form:

The fingertips have a particularly high density of receptors, which is why two touching points can be told apart there at a distance of only a few millimetres, much less than on the back or the forearm.

Active touch and haptic exploration

Placing an object on a still hand tells us surprisingly little. Moving the fingers over it reveals much more. Psychologists describe typical exploratory procedures: pressing to judge hardness, lateral rubbing to feel texture, enclosing the object to judge its global shape and holding it to estimate weight.

Researchers study these processes with experiments such as haptic search, in which people look for a target, for example a hard sphere among soft ones, using only touch. Such studies show that some properties, like a clear difference in hardness, stand out almost immediately, while others require a slow, systematic search. They also show how strongly perception depends on the way the hand moves, which links touch to the movement patterns described on our page on hand synergies.

Tactile sensors in robot hands

Robots need similar information. Common tactile sensing technologies include:

• Resistive and capacitive arrays that measure the pressure distribution over a fingertip or palm, like a low-resolution image of the contact.
• Piezoelectric sensors that react to fast changes and vibration, useful for detecting the first moment of contact or the onset of slip.
• Optical tactile sensors that film the deformation of a soft skin from inside with a small camera and calculate contact shape and forces from the images.
• Force and torque sensors in the wrist or finger joints that measure the total load.

Combined with soft fingers, as described in our explainer on soft and underactuated robotic hands, these sensors let a robot detect slip and tighten its grip just enough. That is the difference between holding a strawberry and squashing it.

Haptic feedback: feeling through a machine

Haptic technology also works in the other direction, delivering touch to a person. Examples include:

• Teleoperation. An operator steers a remote robot, for instance for maintenance in hazardous environments, and feels contact forces through a force-feedback joystick or exoskeleton.
• Medical training simulators that let trainees feel tissue resistance before they work on patients.
• Wearable haptics such as vibration motors or fingertip devices that press or stretch the skin to signal contact.
• Everyday devices like the vibration patterns in phones and game controllers, a very simple form of haptic feedback.

A key challenge is that touch is fast and rich. A simple vibration can tell you that contact happened, but not how hard or where. Research devices therefore try to deliver several cues at once, or to use a few well-chosen signals that the brain can learn to interpret.

Touch for prosthetic hands

For people using a prosthetic hand, missing touch is one of the biggest limitations. They have to watch the hand to know whether it has closed on an object. Research projects test ways of restoring some feedback, from vibration on the skin of the residual limb to implanted electrodes that stimulate nerves. More about current prosthetic hands and what to expect from them is in our guide to bionic and myoelectric prosthetic hands.

Summary

Touch combines skin sensations with information from muscles and joints, and it works best when the hand moves. Robots copy parts of it with tactile and force sensors, and haptic devices bring touch back to people who operate machines or wear prostheses. For how this fits into the bigger picture of grasping, see the overview of robot hands and touch.