Amplitude variation across bodies needs calibration

What it is

Individual amplitude calibration means recognition thresholds (amplitude, angle, speed) are personalised against the player's own body rather than applied as one fixed set of values for everyone. Human height, arm length, shoulder width, and movement habits vary enormously—the same "raise your arms overhead" gesture corresponds to absolute heights differing by more than thirty centimetres between players 1.5 m and 1.9 m tall, so a fixed threshold is inevitably too strict for some and too loose for others.

Why it happens

A fixed threshold silently assumes all players share one body—an assumption that holds only in statistical averages. What the recognition system measures are absolute quantities in sensor coordinate space (joint angles, end-effector positions, movement speeds), while the gesture's intent is a quantity relative to the player's own body ("above your head," "arms fully extended"). A conversion sits between the two, and the conversion factor is body size. Without calibration, that factor is implicitly set to some average build—smaller-than-average players perform the full gesture and still fall short of threshold (undetected), while larger-than-average players cross it with a minimal movement (over-loose detection). Both groups experience the same system as two different kinds of broken.

Where it stops holding

Calibration is not a universal fix. Some gestures are defined independently of body proportions ("press palms together," "push palm forward"), where a fixed definition is more accurate and personal calibration would only inject error. The calibration flow itself has a cost: every extra setup step is a churn opportunity, so it must stay under ten seconds and use the minimum gesture set needed to derive a height-and-reach ratio (one or two movements). Calibration quality also depends on player cooperation—a careless calibration baseline is worse than none, so the flow needs a consistency check (repeat once to confirm) and should allow recalibration during play.

Applying it

• At session start, capture a body-proportion baseline with one or two reference gestures (lateral arm raise, overhead reach), and convert later recognition thresholds to personal relative values.
• In recognition-failure messages, express amplitude references against the player's own body ("raise your hand to above your head") rather than absolute distances.
• Verification: compare recognition success rates across height groups between fixed and calibrated thresholds. Effective calibration should equalise the groups; if one group still lags, check whether the calibration gestures suit their body proportions.

Related

• Same group: W7.04.1 Static diagrams cannot convey movement · W7.04.2 Teaching needs demonstration and live correction
• Nearby: W7.01 Motion recognition reliability · C1.02 Anthropometric variation · K3.02 Motion controllers
• Search terms: personalized calibration · anthropometric variation · threshold scaling · gesture personalization