Skip to content
Electronics & Semiconductor

Touch Input for AR Glasses: Why the Display Is the Wrong Place

aadmin
The short answer

AR glasses have an input problem that is easy to underestimate: the surface the user wants to touch is not the surface they can comfortably see. The image arrives through a combiner inside the optical path, where adding a sensor means adding a bonded layer, complicating alignment and reducing the light that reaches the eye. The surfaces that do work are the temples, the frame edges and the hinge area, all of which a user can find without looking. Where a sensor genuinely has to sit near the image, on-cell integration decided at module level is the option that keeps the optical stack thinnest, and that decision has to be made long before the frame design is released.

Key takeawaysTouchscreens detect the position of a contact on a surface, and the different technologies trade optical impact against robustness in different ways [1]. · Capacitive sensing works through a sealed cover without moving parts, which suits eyewear that has to survive rain, sweat and cleaning [2]. · In augmented reality the display path is shared with the view of the world, so any layer added there affects both the virtual image and the transmitted scene [3]. · Frame and temple sensors can be operated by feel, which removes the need for the user to keep a cursor in view and reduces the demand on the display. · On-cell sensing integrated during module manufacture avoids a separate bonded sensor sheet, but it has to be agreed when the tail and connector are specified.

Glasses are a familiar object, and that familiarity sets an expectation: the user should be able to adjust or control them without taking them off and without looking. A touchscreen embedded in the display violates that expectation, because the user has to look at the image to aim a finger at it, and in a see-through device the image moves with the head.

The optical argument reinforces the ergonomic one. A combiner-based system balances the light from the display against the light from the scene, and every additional surface in the display path removes some of the former while potentially adding reflections. A touch layer bonded onto the display or onto the combiner therefore has a cost measured in image quality, not only in millimetres.

That leaves the frame. Temples, hinge housings and bridge areas offer surfaces that are mechanically robust, easy to seal and easy to locate by touch. They also sit away from the display link, which avoids the coupling problems that arise when a sensing electrode shares space with a high-speed differential pair.

The rest of this article compares the realistic options and describes how to test them once the device is actually being worn.

Input surfaces that work while the device is worn

A user wearing glasses can reliably find the temple above the ear and the hinge area without looking. Both are large enough for a finger, both are naturally touched when the glasses are put on or adjusted, and both are mechanically connected to the frame rather than to the optical assembly. Sensing at those points therefore does not disturb alignment between the panel and the combiner.

A temple strip also supports a richer vocabulary than it first appears. A single tap, a double tap, a forward or backward swipe and a long press cover most of what a lightweight interface needs, and each can be confirmed with a short haptic pulse so that the user knows the gesture registered without needing visual feedback.

There is an argument for combining a small number of physical controls with sensing. A single mechanical button that wakes the device and a strip that handles navigation is often more reliable in the cold or with gloves than a purely capacitive surface, and the button doubles as a power switch. Where the interaction really must happen on the image itself, a 0.72" 1920x1200 Micro OLED display can carry an on-cell sensor, but that decision has to be taken with the module rather than added to the frame afterwards.

Where to put the sensor on AR glasses

OptionStrengthCost or risk
Temple stripBlind operation, no optical impact, straightforward sealingCurved sensing surface and long-term exposure to sweat and moisture
Hinge housingClear physical landmark and a natural place for a small electrodeLimited area, and the mechanism has to tolerate repeated contact
Frame frontSupports deliberate pointing gesturesLarger sensing area increases accidental activation, particularly in crowds
On-cell on the displayDirect manipulation of displayed itemsAdds a layer to the display path and complicates alignment and repair
External controller or ringFull input vocabulary with precise pointingAn additional object to carry, charge and pair
Illustration: Where to put the sensor on AR Decorative illustration for the section "Where to put the sensor on AR"; visual only, carries no data.

Testing input on a device that is actually worn

A sensor that works on a fixture often behaves differently on a head. The frame warms against the skin, moisture accumulates, and the electrical environment changes as the device's own electronics run. Capacitive baselines drift under these conditions, so a controller that recalibrates quickly will interpret drift as a touch and produce phantom activations.

The practical mitigations are the same ones used in other wearables. Track the baseline slowly so that drift is separated from a deliberate contact, require a minimum contact duration before an action is executed, and provide confirmation feedback so that a missed gesture is immediately apparent. A design that assumes every contact is intentional will feel unreliable even when the electronics are working correctly.

Testing should include the awkward cases: a finger that is slightly damp, a frame that has been worn for an hour, a cold morning where the user is wearing gloves, and the position of the hand while walking. Those four conditions account for most of the complaints that appear once a device reaches users.

If the sensor has to be on the display path

On-cell integration at module level avoids a separate bonded sheet and keeps the stack thinner, but it commits the project to a custom module. That decision changes the tail, the connector and the test plan, so it belongs in the same review as the panel selection rather than in a later mechanical revision.

Feedback without a screen

Because the eyes are occupied with the world, confirmation should not depend on the display. A short vibration, a click from a small actuator or an audible tone gives the user certainty that the gesture registered, and it removes the temptation to repeat a gesture that did work.

A useful constraint for AR input design: if a gesture cannot be performed and confirmed while walking down a street, it will not be used. That test eliminates most screen-based interaction immediately and points the design at the frame.

Illustration: A useful constraint for AR input Decorative illustration for the section "A useful constraint for AR input"; visual only, carries no data.

Sources

  1. Wikipedia: Touchscreen
  2. Wikipedia: Capacitive sensing
  3. Wikipedia: Augmented reality

Frequently asked questions

Can the combiner itself be used as a touch surface?

It can be fitted with sensing, but the combiner sits in the optical path where any added layer affects both the virtual image and the transmitted view. The mechanical and optical cost is usually higher than placing the sensor on the frame, where it can be located by feel.

What is on-cell sensing and when is it worth it?

On-cell sensing builds the electrode pattern onto the display substrate during module manufacture, removing the separate bonded sensor sheet. It is worth considering when interaction must happen directly on the image, and it has to be agreed when the module tail and connector are specified.

How can phantom touches be reduced?

Track the capacitive baseline slowly so thermal drift is not mistaken for contact, require a minimum contact duration for actions that change state, and confirm each accepted gesture with haptic or audible feedback.

Should AR glasses include physical buttons?

A small number of mechanical controls is often a good complement to sensing, because they work with gloves and in the cold and can wake the device without continuous scanning. One wake button plus a sensing strip is a common and robust combination.

Find the shoe that actually fits your foot.

60-second GaitMap scan. 60-day fit guarantee. Free return shipping both ways — no restocking fee.

Find My Shoe →