Simulation of Motion

I attempted to simulate this effect with SPICE, but once again convergence problems were encountered. Therefore, I used a modified version of my program to perform the transient analysis. For transient analysis, I assumed that the motion was slow enough that it could be assumed that the network attained a stable state between each change in the input. This enabled me to simply run the simulation over and over again, with slightly different input signals, with the previous output as the initial state of the network. The results of this are shown in Figure 8.3. This input signal is shown as dotted lines. The original network fails to segment once the edge moves. The modified network segments on either side of the moving edge. As a result the modified network's response is exactly equal to the input.

Figure 8.3: Response of New and Old Networks to Motion.