Chapter 2. Prehension

Chapter 7 - Opposition Space Phases 293
be within; i.e., as long as the applied force of the fingers is directed
within the cone of friction, the object will not slide relative to the
fingers. The fingers do not have to make contact simultaneously,
since knowledge of task mechanics will have ensured that being in the
ballpark of the chosen opposition vector will rotate or push the object
into the grasp, instead of out of the grasp.
Two major goals seem to be at work, which in their own way
influence both the arm and the hand. Firstly, perception of the
location of the object influences movement parameters; uncertainty in
object location dictates slowing down in the vicinity of the object,
particularly if the objective is not to bowl it over or crush it. This
affects both the transport and orientation of the palm (change in
velocity profile, systematic changes in palm orientation) and the
grasping component (open hand wider before contact is anticipated).
Secondly, perception of force-related object properties (e.g., weight,
surface texture, surface sizes) and goals for task performance (e.g.,
direction and type of motions to impart, forces to apply) show
systematic kinematic and kinetic effects appropriate for the upcoming
grasping demands. The hand must be postured to optimize the force
generating muscles for the task.
Different models of control have been suggested. The level of
these commands could initially be in hand space, thus a trajectory
would be generated involving hand positions and speeds along a path
in a body or world coordinate frame. The alternative is to do this in
joint space, so that paths are specified in an intrinsic frame of
reference, such as shoulder and elbow joint angles. The actual
generation of a motor command would involve translating these
commands (or even high level goals) into the space within which the
actual movement is being driven. Models have suggested how a
motor command could occur at the joint angle level, joint torque level,
and muscle level. Experimental evidence shows the computation of a
population vector for direction of movement. If the control variables
are kinematic ones, this is an inverse kinematic problem. If the control
variables are dynamic ones, such as joint torques over time, this is an
inverse dynamic problem. Inverse problems are underconstrained in
that there are non-unique solutions, especially in a system as
complicated as the human arm, involving 11 degrees of freedom, over
30 muscles, thousands of muscle fibers, multi-joint muscles, and
coupled degrees of freedom. For limiting the computation towards
selecting a unique solution, cost functions and/or constraint
satisfaction networks have been suggested that minimize some
measure, such as potential energy, or movement time, or changes in