Choosing a reference frame between head pointing and body orientation
Aliases: egocentric ray · body-relative pointing · world-locked head gaze
What it is
A head-controlled ray needs a reference frame: world, trunk, or the current view plane. The same physical head turn sweeps a wall menu in world coordinates, may be only a small yaw relative to the chest in a trunk frame, and is cursor motion in the headset image in a view frame. The wrong frame feels like “I did not turn my head and the pointer moved” or “I turned my body and the pointer stayed nailed down.” This is a coordinate problem, not a precision problem and not a neck-pain problem.
Why it happens
Head-orientation sensors report pose against some anchor. Anchored to the room (world), body rotation sweeps the ray across world objects—right for real furniture or spatial anchors. Anchored to the trunk, a turn carries the whole interface with the chest and the ray relative to the sternum does not change—right for a panel that “sticks in front” while sitting in a swivel chair. Anchored to the headset screen, head direction does not move the cursor at all; that is no longer head pointing but a screen-fixed UI.
Conflict appears in mixtures: a world button and a body-locked panel at once, and the user cannot tell whether this head turn selects a lamp in the room or an item on the panel. On a vehicle or vessel, a world frame writes carrier motion as input. Lying down or on one side, the trunk’s “up” no longer matches gravity, and pitch defined against gravity flips onto an unintelligible axis.
Studying it
Run the same head-pointing tasks under three anchors: point at a physical object in the room, point at a body-locked panel, turn on a swivel chair and point again. Outcomes: which class of target was changed by mistake, reorientation time, verbal reports of “who does the pointer follow.” A moving platform (chair, treadmill, simulated cockpit) is necessary; a static lab makes world and trunk look alike. Trunk heading cannot be estimated from the headset alone, or the trunk frame has no independent measure.
Where it stops holding
In a fully static app whose interface exists only inside the headset image, the choice of frame is almost invisible because the three anchors collapse to the same picture coordinates. The moment spatial anchors, co-located other people, or walking appear, the choice is obvious. People with vestibular disease or visual–vestibular conflict find a world-frame pointer that rides body rotation nauseating. For wheelchair users the “trunk” is fused with the seat, the trunk frame is actually a seat frame, and turning is constrained by the chair.
Applying it
- Mark an anchor for each class of target: room objects in world, worn panels on the trunk, and give a brief frame cue on a switch.
- Do not default to uncompensated world-frame head pointing in vehicle, vessel, or walking settings.
- Verify by having users turn then point, checking that room targets and worn panels move or stay as intended, and collecting “who does the pointer follow” reports.