timing window
Undergraduate Category: Physical and Life Sciences Degree Level: BS in Physics Abstract ID #772
Want to Win the World Series? Here’s How to Pitch the Perfect Ninth. Dena Guo , Meghan E. Huber , Dagmar Sternad 1
1,3,4
Physics, Bioengineering, Biology, Electrical and Computer Engineering 2
3
4
RESULTS
ABSTRACT
In a historic moment for the Red Sox, pitcher Koji Uehara clinched the 2013 World Series by pitching a perfect ninth inning. Pitching is a skill that not only requires total body coordination, but also releasing the ball at just the right moment — or so it appears. Such accuracy is remarkable given the complex human neuromotor system and the numerous noise sources that lead to performance variability. How can one consistently execute tasks that require accurate timing with our noisy neuromotor system? We hypothesized that humans can only achieve high timing accuracy by creating timing windows in their arm trajectory, during which all ball releases result in hitting the target. The larger the timing window, the less variability in release timing affects throwing performance. In this experiment, we examined how humans learned a virtual throwing task over 15 days of practice. Participants rotated a lever arm about the elbow and released a virtual ball to hit a target. While participants decreased their timing error, they reached a limit at 13ms after 5 practice days. Consistent with our hypothesis, participants continued to improve by shaping their arm movements to increase the timing window for release. At the end of practice, it was the participants with the largest timing windows, not the participants with lower timing error, who had the best performance. These findings revealed that the key to throwing a game-winning pitch lies in the shape of the arm trajectory instead of the accuracy in timing the ball release.
THROWING ERROR
Throwing error is a measure of performance. It is defined by the absolute distance between the center of the target and the trajectory at its closest point.
THROWING ERROR (cm)
BACKGROUND AND METHODS
11 right-handed subjects | 240 throws a day | performance tracked over 11 days Target
TASK
Center Post
Error
Poteniometer
Skittles is a game often played in British pubs. Players swing the ball tethered to a post around the post in order to hit a target “skittle” on the other side.
HYPOTHESIS
MODEL TASK
The mechanics of Skittles is modeled as a two-dimension spring system. The ball (white circle) is attached to the center post (red circle), which is the origin, by two orthogonal springs. The ball’s trajectory (dotted line) depends on two variables: the release (Φr) and the release velocity (vr).
EXPERIMENTAL TASK
Subjects manipulate a metal lever arm by rotating their forearms about the elbow. They throw the ball by pressing and lifting their index finger from a pressure switch. A top down view of the task is projected on a screen in front of them with the center post (red circle), target skittle (yellow circle), and ball (white circle).
We hypothesized that humans can only achieve high timing accuracy by creating timing windows in their arm trajectories, during which all ball releases result in hitting the target.
STATE SPACE −500
E
SE
N
−200 140
SI
The error of each throw is completely determined by the ball’s velocity and angle at the time it is released. The solution manifold (thin black line) represents all pairs of release velocities and angles that would result in the ball going through the center of the target (i.e. zero error). The time-sensitive trajectory requires accurate timing to result in a target hit.
TI
T
V TI
SE ME NS IT IV E
IM
IN
RELEASE VELOCITY
E
ERROR (CM)
SOL MANUTION IFO LD
100
RELEASE ANGLE
60 0
TIMING WINDOW
standard deviation
8 7 6 5 4 3 2 1
1
2
3
4
5
6
DAY
7
8
9
10
11
mean standard deviation
20 18 16 14 12 1
2
3
4
5
6
DAY
7
8
9
10
11
22
TIMING WINDOW
PC
2D Model
9
10
Switch
Ball
REAL TASK
Timing error is defined by the difference in time between the ideal release and the actual release.
The time-insensitive trajectory creates a large timing window, where the ball can be released at any time and still result in a target hit.
The timing window is determined by the length of time the trajectory spends along the solution manifold.
CORRELATIONS 3.5 3 2.5 2 10
12
14
16
18
20
22
24
20 18 16 14 12 10 8
mean
6
standard deviation
41
2
3
4
5
6
7
8
9
10
11
DAY
4
1.5
TIMING WINDOW (ms)
METHODS
TIMING ERROR
THROWING ERROR (ms)
AIM
mean
22
TIMING ERROR (ms)
The goal of this experiment was to observe how subjects learned a novel throwing task and to probe the limits of human timing precision. Are MLB pitchers good because they have very accurate timing or is there something else at play?
10
26
28
TIMING WINDOW (ms)
On day 11, subjects with a higher timing window had lower throwing error. (p = 0.0032, R2 = 0.64)
THROWING ERROR (cm)
1
2
4 3.5 3 2.5 2 1.5 10
11
12
13
14
TIMING ERROR (ms)
15
16
On day 11, subjects with a lower timing error did not necessarily have a lower throwing error (p = 0.75, R2 = 0.01)
CONCLUSION
Throwing error decreases over time — subjects perform better at the task. Timing accuracy plateaus around 13 ms after day 5.
ACKNOWLEDGEMENTS
Dena Guo: NEU Provost Grant Meghan Huber: NEU Graduate College of Engineering and The Mathworks Dagmar Sternad: NIH R01-HD04563, NSF DMS-0928587, AHA 11SDG7270001, and NEU Provost Tier 1 grant
However, the length of the timing window continues to increase throughout day 11.
Timing window is more important than timing accuracy for consistent performance.
Want to Win the World Series? Here’s How to Pitch the Perfect Ninth. Dena Guo , Meghan E. Huber , Dagmar Sternad 1
1,3,4
Physics, Bioengineering, Biology, Electrical and Computer Engineering 2
3
4
RESULTS
ABSTRACT
In a historic moment for the Red Sox, pitcher Koji Uehara clinched the 2013 World Series by pitching a perfect ninth inning. Pitching is a skill that not only requires total body coordination, but also releasing the ball at just the right moment — or so it appears. Such accuracy is remarkable given the complex human neuromotor system and the numerous noise sources that lead to performance variability. How can one consistently execute tasks that require accurate timing with our noisy neuromotor system? We hypothesized that humans can only achieve high timing accuracy by creating timing windows in their arm trajectory, during which all ball releases result in hitting the target. The larger the timing window, the less variability in release timing affects throwing performance. In this experiment, we examined how humans learned a virtual throwing task over 15 days of practice. Participants rotated a lever arm about the elbow and released a virtual ball to hit a target. While participants decreased their timing error, they reached a limit at 13ms after 5 practice days. Consistent with our hypothesis, participants continued to improve by shaping their arm movements to increase the timing window for release. At the end of practice, it was the participants with the largest timing windows, not the participants with lower timing error, who had the best performance. These findings revealed that the key to throwing a game-winning pitch lies in the shape of the arm trajectory instead of the accuracy in timing the ball release.
THROWING ERROR
Throwing error is a measure of performance. It is defined by the absolute distance between the center of the target and the trajectory at its closest point.
THROWING ERROR (cm)
BACKGROUND AND METHODS
11 right-handed subjects | 240 throws a day | performance tracked over 11 days Target
TASK
Center Post
Error
Poteniometer
Skittles is a game often played in British pubs. Players swing the ball tethered to a post around the post in order to hit a target “skittle” on the other side.
HYPOTHESIS
MODEL TASK
The mechanics of Skittles is modeled as a two-dimension spring system. The ball (white circle) is attached to the center post (red circle), which is the origin, by two orthogonal springs. The ball’s trajectory (dotted line) depends on two variables: the release (Φr) and the release velocity (vr).
EXPERIMENTAL TASK
Subjects manipulate a metal lever arm by rotating their forearms about the elbow. They throw the ball by pressing and lifting their index finger from a pressure switch. A top down view of the task is projected on a screen in front of them with the center post (red circle), target skittle (yellow circle), and ball (white circle).
We hypothesized that humans can only achieve high timing accuracy by creating timing windows in their arm trajectories, during which all ball releases result in hitting the target.
STATE SPACE −500
E
SE
N
−200 140
SI
The error of each throw is completely determined by the ball’s velocity and angle at the time it is released. The solution manifold (thin black line) represents all pairs of release velocities and angles that would result in the ball going through the center of the target (i.e. zero error). The time-sensitive trajectory requires accurate timing to result in a target hit.
TI
T
V TI
SE ME NS IT IV E
IM
IN
RELEASE VELOCITY
E
ERROR (CM)
SOL MANUTION IFO LD
100
RELEASE ANGLE
60 0
TIMING WINDOW
standard deviation
8 7 6 5 4 3 2 1
1
2
3
4
5
6
DAY
7
8
9
10
11
mean standard deviation
20 18 16 14 12 1
2
3
4
5
6
DAY
7
8
9
10
11
22
TIMING WINDOW
PC
2D Model
9
10
Switch
Ball
REAL TASK
Timing error is defined by the difference in time between the ideal release and the actual release.
The time-insensitive trajectory creates a large timing window, where the ball can be released at any time and still result in a target hit.
The timing window is determined by the length of time the trajectory spends along the solution manifold.
CORRELATIONS 3.5 3 2.5 2 10
12
14
16
18
20
22
24
20 18 16 14 12 10 8
mean
6
standard deviation
41
2
3
4
5
6
7
8
9
10
11
DAY
4
1.5
TIMING WINDOW (ms)
METHODS
TIMING ERROR
THROWING ERROR (ms)
AIM
mean
22
TIMING ERROR (ms)
The goal of this experiment was to observe how subjects learned a novel throwing task and to probe the limits of human timing precision. Are MLB pitchers good because they have very accurate timing or is there something else at play?
10
26
28
TIMING WINDOW (ms)
On day 11, subjects with a higher timing window had lower throwing error. (p = 0.0032, R2 = 0.64)
THROWING ERROR (cm)
1
2
4 3.5 3 2.5 2 1.5 10
11
12
13
14
TIMING ERROR (ms)
15
16
On day 11, subjects with a lower timing error did not necessarily have a lower throwing error (p = 0.75, R2 = 0.01)
CONCLUSION
Throwing error decreases over time — subjects perform better at the task. Timing accuracy plateaus around 13 ms after day 5.
ACKNOWLEDGEMENTS
Dena Guo: NEU Provost Grant Meghan Huber: NEU Graduate College of Engineering and The Mathworks Dagmar Sternad: NIH R01-HD04563, NSF DMS-0928587, AHA 11SDG7270001, and NEU Provost Tier 1 grant
However, the length of the timing window continues to increase throughout day 11.
Timing window is more important than timing accuracy for consistent performance.
