The pattern-association methods from chapter 4 and 5 are applied to a robot arm, which is equipped with a camera. An object can be grasped by associating the object's image with an arm posture. Arm postures can be recalled without sensory feedback (open-loop). This enables the robot to perceive (to understand) the position and orientation of an object by associating an appropriate grasping posture and not by mapping the image coordinates to the three-dimensional space. Training data were collected by random exploration. With image preprocessing steps comparable to functions that are known to be performed by the primary visual cortex, the dimensionality of the original images was reduced. On the other hand, the dimensionality of the posture, defined by six joint angles, was increased. This increase improved the performance.