Motor input

In this experiment, the robot movements are limited to 7 possible rotations from $-90^\circ$ to $90^\circ$, with $30^\circ$ steps. The motor layer is composed of $N^{(3)}=7$ neurons, each associated with one motor command, so that the neuron with maximal activation determines the movement. The robot is thus limited to rotations, in order to allow simple correspondences between the motor movements and the visual field.

Learning is done with a forcing motor signal on layer 3. The purpose is to link (associate) this motor flow with the incoming visual flow. Like in previous simulations, we use a periodic signal. This motor signal corresponds to the 3-periodic sequence of rotations $(+30^\circ , +60^\circ , +90^\circ )$, so that after 3 steps, the robot has made a half-turn, and after 6 steps, the robot faces its initial visual field. Apart from visual noise and small angular shifts, the visual input signal is thus supposed to be of period 6 (Fig.9).

Figure 9: Successive positions of the robot after a motor command sequence ($+30^\circ$,$+60^\circ$,$+90^\circ$). The robot stands in an open environment (it is not a simulation). The association of a set of landmarks (high curvature points, denoted as 'x' on the figure) with their angular positions constitutes the visual input. After 3 commands, the robot is facing backwards (issuing these commands again let the robot face its initial visual scene).