Hi, Krumlink.
I'm mystified as to why he uses six photosensors. I have read a few discussion about designs of line following robots. General concensus usually says only two sensors are needed, and more may just add to design problems. A lot of these designs consist of just two sensors, two transistors, batteries and two motors. That is, one motor per wheel. Each sensor drives a transistor which drives a wheel motor. As long as each sensor detects reflected light, both transistors are on, and both wheels are driven. When the robot wanders across (into?) the line (we're talking black line/white floor), the sensor stops conducting. This leads to the transistor "pinching off" or going open circuit, which limits or kills the wheel motor. Thus, the still driven wheel forces the robot to turn towards the now "dead" side. This brings the sensor back over the floor, so it starts to conduct again. This turns the transistor back on, which now drives the wheel motor,and we're running staright again. Or we overshoot, the other sensor now goes dark, its motor dies... If the design doesn't allow for controlled, gradual speed up/down of the wheel motors, the robot ends up lurching along drunkenly, side to side. You can probably account for this with an RC circuit on each transistor's base. When the sensor "goes dark", the RC circuit bleeds off slowly, thus eliminating completely killing the motor and avoiding the "lurch".
There's lots of this on the web, as you already know. This is the minimalist approach. Of course processor controlled contraptions are the only way to go.
Later!
kenjj