17.01.2013 Views

Chapter 2. Prehension

Chapter 2. Prehension

Chapter 2. Prehension

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Chapter</strong> 5 - Movement Before Contact 115<br />

1989), mean-square error of jerk3 (Flash and Hogan, 1985), and<br />

mean-square error of torque-change (Uno, Kawato, Suzuki, 1989). A<br />

cost function constrains the potential hand paths and velocities, in that<br />

the CNS makes a selection out of the infinite number of possible by<br />

minimizing this measure. One such model is the ‘minimum jerk<br />

model’ (Flash and Hogan 1985), which suggests that the square of the<br />

jerk of hand position integrated over the entire movement is<br />

minimized. Flash and Hogan showed that the trajectory uniquely<br />

defined by this criterion function agrees with experimental data in<br />

various regions of the workspace4.<br />

In summary, trajectory planning involves transporting the hand to<br />

a location by a controller. Cortical areas, such as the motor and<br />

somatosensory cortex, are critical for neural control. Sensory<br />

information is provided by exteroceptors and/or proprioceptors. How<br />

the CNS transforms a goal into motor commands is a critical question,<br />

particularly in light of the fact that so many degrees of freedom exist at<br />

the joint level, the muscle level, and particularly at the muscle fiber<br />

level. Exacerbating this are problems such as motor equivalency,<br />

coupled degrees of freedom, and an endless number of solutions.<br />

Cost functions have been proposed as a way to constrain the potential<br />

number of solutions.<br />

5.2 Transforming a Goal into Motor Commands<br />

Goals must be transformed into motor commands. If the goal is<br />

specified in some extrinsic coordinate reference frame, the CNS could<br />

possibly compute an arm trajectory also in an extrinsic coordinate<br />

frame. This trajectory could then be transformed into an intrinsic<br />

coordinate frame (e.g., joint angles), and then into intrinsic dynamics<br />

(e.g., muscle activity). Along this hierarchical scheme and coming<br />

from a robotics framework, Uno, Kawato and Suzuki (1989) put forth<br />

a computational model for voluntary movement as seen in Figure 5.3.<br />

Starting from the environmentally-defined goal (e.g., target location,<br />

object size), they proposed the following computations:<br />

1) determination of desired traectorv - the goal of the movement,<br />

such as picking up a coffee mug in order to drink from it, is<br />

translated into an extrinsic kinematic coordinate frame for a path<br />

3Jerk is the rate of change of acceleration.<br />

4The workmace is a collection of points to which the hand of an arm can reach.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!