Page 144 - Concise Encyclopedia of Robotics

Grasping Planning
paths are defined as sets of points at the greatest possible distances from
obstacles, barriers, or hazards. In a hallway, for example, the path goes
down the middle. The same is true as the robot passes through doorways.
The paths in other places depend on the locations of the nodes, and the
arrangement of obstructions in the rooms or open areas (Fig. 2).
See also COMPUTER MAP. Compare METRIC PATH PLANNING and TOPOLOGICAL PATH
PLANNING.

Graphical path planning—Fig. 2

GRASPING PLANNING
Grasping planning refers to the scheme that a robot arm and gripper use
to get hold of a chosen object.
Suppose a person tells a robot to go to the kitchen and get a spoon.
The robot uses gross motion planning to find the kitchen, and fine motion
planning to locate the correct drawer and determine which objects in the
drawer are the spoons. Then the gripper must grasp a spoon, preferably
by the handle rather than by the eating end. The robot must not get a fork,
or two spoons, or a spoon along with something else such as a can opener.
Hopefully, the silverware is arranged logically in the drawer, so spoons
are not randomly mixed up with forks, knives, can openers, and other
utensils. This can be ensured by programming, as long as the robot (but
only the robot) has access to the drawer. If there are children in the
household, and if they get into the silverware drawer, the robot had better
be able to cope with mixed-up utensils. Then, getting a spoon becomes a
form of bin picking problem.
Close-up, detailed machine vision, such as an eye-in-hand system, can
ensure that the gripper gets the right utensil in the right way. Tactile sensing
might also be used, because a spoon “feels” different than any other kind
of utensil.
See also BIN PICKING PROBLEM, EYE-IN-HAND SYSTEM, FINE MOTION PLANNING, GROSS MO-
TION PLANNING, OBJECT RECOGNITION, and TACTILE SENSING.

