Line Maze Sensor Array Configuration - Possibilities?

I recently did a line maze robot that keeps getting confused at intersections.
If the 5-sensor, in-line array approaches an intersection, say a “T”, and it is even a teeny bit off, one sensor doesn’t see black (should be 11111 and is 01111) - - well 01111 is also legal for a right-turn only intersection, so the in-line array would occasionally lie to the robot.

So, I’m thinking a different configuration would be more reliable. For example:
The 3pi has a circular arc of sensors like:
…#…
…#…#…
…#…#…
#…#
(Not a good drawing, but picture the #'s as sensors in a smooth circular pattern.)

I also thought that 3 central sensors to detect the line, and two “outside” sensors to sense the upcoming intersection will also work. Something like:
…#…#… (out front to sense 90 degree turn or “T” intersection)

…#…#…#…(Use these to stay on the line)

Has anybody experimented or tried any different configurations??

Hello,

I’ve done a line maze robot with the sensors arranged in a line, and that worked fine. I’m generally skeptical that different arrangements of sensors will get you much. For instance, wouldn’t your proposed arrangement work the same way even if the middle three sensors were moved up to the same height as the outer two sensors? For me, the general approach is to get the sensors as far out as mechanically practical so that you see things coming as soon as possible. Once you have side sensors out far ahead, moving the middle ones in is just unnecessarily limiting your ability to look ahead.

This is just my intuition, so it might be wrong.

- Jan