Large gestures hit furniture and people the wearer cannot see

What it is

A bowling swing is clean in the headset; in the room the coffee table sits on the arc. An unseen collision is furniture and people inside the volume a large gesture sweeps, whom the wearer currently cannot see. Once the eyes are covered, the hands' workspace and the eyes' workspace are no longer the same room.

A guardian grid is usually drawn before the feet meet a wall. It does not enclose a swing. The hand arrives first.

Why it happens

A fully occluding headset removes the room from view; video see-through still offers only a forward, delayed, narrow window. Large input — a racket swing, a throw, a block, both arms opening to reach — sweeps a sector about a metre in radius from the shoulder. Tables, lamps, pets, and people walking past often sit in that sector and out of the current picture. Force is aimed at the virtual target; the arm leaves on a path that is right for the virtual world and wrong for the room.

Boundary systems alarm on trunk or head pose and are late for a swing already in flight. The tracking volume also rewards making the motion large: small motions fall out of frame and out of track, large ones feel coupled, so the input itself pushes the swept volume into the furniture. The collision is on the hand or the controller; pain and damage are on the real side, and the virtual side may log nothing.

Studying it

Place soft obstacles (foam posts, hanging light balls) in the arm-sweep volume. Compare small taps with large swings, and seeing the real room versus full occlusion. Count near-misses; do not wait for contact.

Independent variables: gesture amplitude, see-through on or off, obstacle azimuth relative to the shoulder (front / side / behind). Dependent variables: collisions, near-misses, contact site (hand / controller / head), whether the motion was arrested before contact.

Side and rear obstacles expose “unseen” better than frontal ones. Frontal space still has picture; people can sometimes stop.

Where it stops holding

Optical see-through still shows the real table, so this leaf is mainly about full occlusion and poor video see-through. Seated tasks that never leave a keyboard-sized volume will not hit the coffee table. An empty room cleared at setup drops the rate; it does not generalise to a living room. Children's shorter arms and less predictable paths make a different sector than an adult sample. A controller adds length; estimate the swept volume at the tip, not the fingertip.

Applying it

• Do not make full-arm swings the default input. Keep primary actions inside a small volume below the elbow and in front of the chest.
• If the content must use large motion, mark solids within an arm on the side and in front — environment understanding or a manual scan — and block the virtual target before the arm goes, rather than after contact.
• Guardian should cover the height and the sides the hands can sweep, not only the wall at the feet.
• How to check: put a soft obstacle an arm to the side and run the swings the content actually uses. Hits or grazes mean input volume and visible room are not aligned — shrink the motion or mark that side as out of bounds, rather than adding a line of “watch your surroundings.”

Related

• Same group: N2.13.2 With vision occluded, estimates of one's own position drift · N2.13.3 Bystanders cannot predict the wearer's next movement · N2.13.4 Banning dangerous actions in the content is more effective than after-the-fact warnings
• Nearby: N4.04 Room Scale and Safety Boundaries · N5.06 Bystanders
• Search terms: unseen collision · playspace · arm-sweep volume