Chapter 2
State of the art
Grasping and manipulation tasks, in general, require a priori knowledge about
the geometric characteristics of the objects and of the scene. One of the first
approaches about grasping in unknown environments can be found in [111], where
a visual control of the grasp is performed employing visual information in order
to track both the positions of the object and of the fingers.
A method to grasp an unknown object using information provided by an algo-
rithm with deformable contour model is proposed in [74]. In [110], an omnidirec-
tional camera is used to recognize the shape of the unknown objet, while grasping
is achieved on the basis of a grasp quality measure, using a soft-fingered hand.
The previous mentioned works are concerning about the general problem of
the grasp in unstructured environments, and in [10] it is clearly stated that the
task of autonomous grasping can be generally divided into:
· detection of the object;
· recognition of the object;
· coarse end-effector alignment;
· preshaping;
· vision-guided grasping;
· execution of the desired grasp action.
Starting from this concept, it is straightforward to recognize that the previous
actions can be grouped into two main tasks which are namely:
· object's recognition/detection;
· grasp planning.
5
Summary :
Chapter 2 State of the art Grasping and manipulation tasks, in general, require a priori knowledge about the geometric characteristics of the objects and of the scene. The previous mentioned works are concerning about the general problem of the grasp in unstructured environments, and in [10] it is clearly stated that the task of autonomous grasping can be generally divided into: · detection of the object;
Tags :
grasp,grasping,object,unknown,information,enironments,recognize,preious,tasks,general,objects,using,isual