Vision
provided by

Light Following Robot using a Neural Network Controller

The robot in this video has two light sensors mounted on the front pointing slightly left and right. The control algorithm uses a simple 4-neuron artificial neural network to map the input light intensities to motor control signals. There is no learning in this video--that will be shown in a later video. The four connections in the neural net used in this video were set ahead of time to fixed values that were known to work.

For more information see:

http://www.pirobot.org/blog/0005/