Hello Michael!
A *very* good question, indeed! I would like to join you and build this model also

A few thoughts:
- There are light barriers at both ends. The carrier itself does not contain a lamp (as is used in the l-e1 manual for a similar model).
- There are only 2 relays with amplifier (h4RB) used. I would guess each of them is controlled by one of the light barriers.
- Of both relays, all 3 contacts of the top switch are used, but only 2 (the working contact) of the lower switch.
- I see 1 button underneath the electronic modules.
- So I guess the top switches of the 2 relays are connected with their center contact to one of the motor poles each, and the working and non-working contacts (on the right of the module) are connected to + and -, resp. Thus, exactly if 1 of the 2 relays is "working", the motor turns, and its direction is determined by which of the 2 relays is "working" (I don't know the correct English wording for that). This is an XOR-like behavior.
- Further, I would guess that the 2 lower switches of the 2 relays are used in some self-hindering circuit. The photo cells are connected to the relays inputs via their lower switch. Thus, when a light barrier is interrupted, the relays falls off, and it stays off even if the wagon has left the light barrier region. The motor continues to turn until the other barrier gets interrupted.
- The button should have some bridging functionality and turn on a relay even if its light barrier is interrupted. This point is my weakest, as I have not yet analyzed how, exactly, this one button could be wired in order to have the correct effect on both relays.
All this is pure theory just from watching the image for some time. I have not yet verified or tested in with real circuitry. I would, however, like to do so as soon as I have got time.
Good luck with your own experiments,
Stefan