In most studies about self-driving cars, only the situation at the highway is explained. The automotive freeway is highly predictable and the cars have to maintain the correct distance to each other. A normal distance sensor which is used by the automotive industry since decades allows to operate a car on the highway with low amount of intervention of a human driver.
The situation is different at the traffic inside a city. The last mile consists of unloading zones, unpredictable pedestrians, parking cars and bicycle who are driving in the wrong direction. It's not the exception, but it's the normal case downtown in a city. In the literature i didn't found a reference for formalizing these situation which would allow self-driving cars to enter the last mile. Are any formal models available which simulates the traffic in the last mile?