Face to Face
Someone stops in front of the robot and looks at it. A pattern is a sequence of components, with rules for each channel.
Sequence
Each step is a component. Time counts from the moment the person stops.
StepComponentTime
1Curious · notice0 – 0.6 s
2Approach · owner0.6 – 2.4 s
3Nod · hello2.4 – 3.0 s
4Idle with drift3.0 s – loop
Motion
Layout and spacing.
RuleValue
LayoutFace the person, head at eye level
SpacingStop 40–60 cm away, never closer than 30 cm
Face
Transition and state.
RuleValue
TransitionEyes widen 100 ms before the body moves
Statesoft → wide → blink, back to soft when idle
Sound
Density and feedback.
RuleValue
DensityOne cue per step, two at most
FeedbackShort rising cue on Nod
Common
Rules every channel follows in this pattern.
- Syncall channels start on the same beat
- Fallbackno face: motion +20% · no speaker: face carries the feedback