Frequent faults on the rear doors of buses are a problem with or without a human driver. Infrequent faults aren't worth much prevention unless the consequences are catastrophic.
In the case of a fault an announcement can be made to the passengers that the bus cannot leave until the rear door is clear. This is already a common occurrence with human drivers; the most frequent cause for a rear door to remain open is passengers accidentally triggering it. Automating this processes is no more complicated than timing how long the door has been open and playing a prerecorded message after an appropriate delay, which is something answering consumer-priced machines from 1989 could do.
If bus was delayed for some reason it could immediately report the problem, summon a replacement, and give passengers instructions and an ETA all before the service tech even finishes reading the alert. Passengers on that particular bus would have to wait for the new bus to arrive, but it wouldn't significantly affect anyone else in the transit system, and again isn't any different than the scenario today when a bus fails while in service with a human driver.
Finally, extrapolating from the coordination failure of human-computer system to a computer-only system is pretty meaningless. The particular failure described was caused because the driver was not physically in position to control the bus, which is simply not possible for a functional bus control computer. If the driver were in position having the rear door interlock release the brakes would constitute normal operation.
Also note that door interlocks are not mandatory (and therefore probably not standard in their operations) and at least the ones I drove with did not require additional human input after the door was closed to release the brake -- they released the brake as soon as the door returned to the locked position.