Skip to main content
. 2016 Jul 15;10:5. doi: 10.3389/fnbot.2016.00005

Figure 2.

Figure 2

The structure of the MTRNN with two stacked context layers. The data flow fed into/out of the I/O layer of the RNN consists of the words, robot vision, and joint angles. Blue-colored parts are used only for learning and explained in detail in Section 3.3.2.