Figure 8.

Autonomous exploration. This composite consists of images taken every 30 s or so over the first hour of the experiment described in section 3.2.1. Although learning has just begun, we already begin to see that the cloud of robot poses is densest (most opaque) near the table. Note that the compositing technique as well as the wide angle lens used here create the illusion that the hands and arms are farther from the table than they really are. In fact, the low arm poses put the hand or the elbow within 2 cm of the table, as shown in Figure 8.