How a strike follows its target: the goal planned at the launch from where
the target is predicted to be at contact, and then online corrections, each
a minimum-jerk unit superposed on the strike's goal, as the target is seen
to move.
A performer observes the target every film frame from the launch. Each
observation predicts the target's point at contact by extrapolating its
current velocity (a first-order prediction, the simplest model of
interception of moving targets; Brenner and Smeets 1996), and the change
of that prediction since the previous observation is added to the goal by a
new minimum-jerk unit that starts one visuomotor correction latency after
the observation and ends at the contact (Flash and Henis 1991: a second unit
superposed on the first, not an aborted and re-planned movement; latency
0.11 s, Prablanc and Martin 1992). Observations stop once a correction could
no longer run for its shortest duration before contact, so a target that
starts moving later than that is not followed: the strike lands on the line
it committed to, which is how a well-timed slip makes a punch miss, and why
a kick that commits at its chamber key (about one latency before impact)
sweeps through where the target was.
A stationary target gives zero corrections and the planned goal unchanged;
a target moving at constant velocity is met exactly, because each
prediction is then exact.
How a strike follows its target: the goal planned at the launch from where the target is predicted to be at contact, and then online corrections, each a minimum-jerk unit superposed on the strike's goal, as the target is seen to move.
A performer observes the target every film frame from the launch. Each observation predicts the target's point at contact by extrapolating its current velocity (a first-order prediction, the simplest model of interception of moving targets; Brenner and Smeets 1996), and the change of that prediction since the previous observation is added to the goal by a new minimum-jerk unit that starts one visuomotor correction latency after the observation and ends at the contact (Flash and Henis 1991: a second unit superposed on the first, not an aborted and re-planned movement; latency 0.11 s, Prablanc and Martin 1992). Observations stop once a correction could no longer run for its shortest duration before contact, so a target that starts moving later than that is not followed: the strike lands on the line it committed to, which is how a well-timed slip makes a punch miss, and why a kick that commits at its chamber key (about one latency before impact) sweeps through where the target was.
A stationary target gives zero corrections and the planned goal unchanged; a target moving at constant velocity is met exactly, because each prediction is then exact.