N2.04.2virtual object penetrationdesignresearch

Without physical resistance, penetration is common

Aliases: hand clipping · unresisted interpenetration · grab-through

What it is

Fingers are already inside the teapot. The porcelain is still there. The skin has not been pushed back. In direct grab, without physical resistance, penetration is common: a virtual object can stop the eye and cannot stop the hand. The hand continues along a path that would have halted in reality, and so it enters the volume.

Penetration is not merely a rendering bug. Rendering can keep the hand drawn outside the surface while proprioception reports that the hand has gone further in. The conflict sits between two senses, not in a missed frame.

Why it happens

A real grasp is braked at contact by a normal force. Muscle spindles and skin together issue a hard stop; further effort becomes squeeze rather than further reach. XR has no such brake. Vision can paint an impassable surface; the arm’s planner still walks through empty space — the planner works in the real room, and the real room is air at that coordinate.

Two visible entries follow. One: the hand is already inside the volume on the way to closing, because closing itself has a component toward the object’s centre. Two: after a supposed “hold,” the hand still drives forward, and the object either comes along or stays put while the hand threads through it. Both share a cause: the stop condition is written in vision, the actuator is flesh that meets no resistance. The more people apply everyday force, the deeper they go — experience here makes it worse.

Freezing the virtual hand on the surface while the real hand continues creates a second conflict: the seen hand stopped, the felt hand did not. Freeze or no freeze, vision and proprioception cannot both be satisfied. Penetration is not a small defect to patch; it is a structural consequence of this input.

Studying it

Have people grab volumetric objects and log real-hand position and virtual-hand position with motion capture; score depth and count of entries into the interior.

Independent variables: object size and material look, whether the virtual hand is clamped to the surface, grasp speed, whether the object is allowed to follow on contact. Dependent variables: peak penetration depth, fraction of grasps that enter, subjective strength of the visual–proprioceptive clash, whether people switch to a lighter, slower fake movement.

Keep “through the surface” apart from “stopping centimetres short.” The first is missing resistance; the second is volume alignment or tracking offset. A single “failure rate” mixes the two. Fast grasps amplify penetration because they shrink the time a brake would have needed.

Where it stops holding

When a fingertip actually meets a desk, a wall, or the other hand, reality returns the resistance and entry is clipped by a physical surface — a virtual object that happens to sit on that desk will suffer less. An exoskeleton or grounded force-feedback device supplies a normal force and the premise drops. Flat, almost zero-thickness cards feel less penetrated than teapots and spheres, because the interior path is short. Making the object snap to the hand on first contact, no longer requiring the hand to stop at the surface, bypasses the need to halt; penetration is avoided, not solved. Slow lab grasps under the instruction “take it gently” underestimate how deep everyday force drives the hand.

Applying it

  • Do not let the visual surface claim “this is a solid shell” while the hand is allowed to hit it at everyday speed. Either accept entry and let the object follow from contact onward, or make grabbables obviously light, thin, and yielding.
  • When the real hand is already inside the volume, do not pin the virtual hand to the surface and pretend nothing happened; the pin makes a more glaring hand–eye split.
  • For objects that must look solid, lower the expected grasp speed: approach, hover, then close — not a swing into the mass.
  • How to check: film the real hand against the virtual object from the side and measure the deepest frame. Repeated entry past the object’s radius at everyday force means missing resistance was not caught by any strategy — not something “draw it more solid” will fix.

Related

  • Same group: N2.04.1 Direct grab matches everyday experience · N2.04.3 It only applies to objects inside reachable space
  • Nearby: N2.09 Direct Grab and Virtual Touch · N2.06 Reach Space
  • Search terms: virtual object penetration · missing reaction force · visual-proprioceptive conflict

Cards in the same group

Quick Actions

Share

Share this page

ios_share

https://hci.top/en/handbook/N2.04.2