automatic selection of point cloud, considering blob dimension and changing robot pose

[giuliavezzani]

This is to find an automatic way for selecting the initial pose of the first arm, considering the quality of the blob

[giuliavezzani]

In order to make point cloud acquisition and filtering automatic, I implemented this approach (to be tested on the robot):

• n poses for the first arm and the head are a priori chosen (currently n=3)
• a minimum size for a good blob is given (I found that we should use min_blob_size=7000)
• for each pose, we acquire a blob, checking if the number of points is enough. As soon as a good blob is acquired, the point cloud is filtered and we move to the next step, without necessarily checking all the chosen poses:

while (blob.size()<min_blob_size)
{
       goToNextPose()
       acquirePoints()
}

 filterPoints()

[giuliavezzani]

The method has been successful tested on the robot (see #28) and a set of initial poses for left arm as first arm have been selected. New tested can be executed for improving this set.