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> 4 - Planning of <strong>Prehension</strong> 81<br />

environments, since it eliminates uncertainty without the use of<br />

sensors. Mason calls operations that eliminate uncertainty funnels,<br />

and argues that manipulation funnels can be used in grasping objects<br />

because they eliminate position uncertainty. The use of a manipulation<br />

funnel is seen in the way humans let a resting pencil rotate into their<br />

grasp by sweeping it up with their fingers.<br />

Is it possible to combine perceived object properties and task in-<br />

formation into a summary representation that the CNS might have ac-<br />

cess to during planning? In Figure 4.7, a process is suggested. For<br />

regularly shaped objects (such as cylinders, blocks and spheres) with<br />

an equal distribution of mass, the specific location on the object to<br />

grasp is not as important as the fact that two (in general) locations are<br />

chosen that are generally parallel to each other so that an opposition<br />

between two virtual fingers can be applied. We call those two surfaces<br />

the ormosable surfaces. If they are not parallel, then due to task<br />

mechanics, the object will be either pushed into or away from the<br />

grasp (see <strong>Chapter</strong> 6). The surfaces have visible characteristics, such<br />

as length, spatial density, and a radius of curvature, all available to the<br />

CNS during the planning process. Klatzky, Lederman, and Reed<br />

(1987) showed that surface spatial density (an aspect of texture) is an<br />

characteristic accessible by vision, as is size. Surface texture relates to<br />

the forces needed (rough surfaces are easier to get a grip on).<br />

Cutkosky and Newell (Cutkosky, 1989 Cutkosky and Howe 1990;<br />

Newell, Scully, Tenenbaum & Hardiman, 1989) demonstrated that<br />

size is thought about in terms relative to the hand. The critical hand<br />

planning parameters about object size are the object width (how wide<br />

does my hand have to open?) and the object length (how many fingers<br />

can I use in VF2?). An opposition vector can be drawn between the<br />

two surfaces with its magnitude being the width of the object between<br />

the two surfaces. Jeannerod and Decety (1990) and Chan et al.( 1990)<br />

demonstrated that object size can be judged in terms of hand opening<br />

size, and Marteniuk et al. (1990) showed that the opening of the hand<br />

relates to object size. For irregularly shaped objects (such as mugs,<br />

flashlights, tools, etc), Klatzky, Lederman, and Reed (1987) have<br />

shown that global shape is accessible to the vision system, thus<br />

making it possible to plan without specific details being known.<br />

Again, knowledge about task mechanics is relevant here, because it<br />

provides a mechanism for anticipating the location of the center of<br />

mass. Handles are easily perceivable aspects of objects, because they<br />

are regularly shaped and in hand-related sizes. As Napier (1980)<br />

pointed out, handles lend themselves to a posture relevant to the task<br />

(e.g., power tasks need power grasps which need larger dimensions

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

Saved successfully!

Ooh no, something went wrong!