Here is a question: how would the future city plan look like? Xiong’an New Area, Hebei, China, has been planned as a future city and smart city. The most important part of the plan is the driver-less cars. Modern city was absolutely shaped by cars. So, how will the smart cars shape our city in the future? With machine learning, we could build a new future city plan.
[An Old Movie]
It’s no doubt a great pleasure to travel freely in the city. But now we have to endure traffic jams.
Is this kind of experience possible in the future cities?
[Research Question]
As the plan said, there will be less private cars in the city. And it would be easy to travel another city fast with High Speed Railway.
The city would be more walking friendly.
Actually, a huge system for driver-less vehicles has been built in Xiong’an.
Different kind of power had shaped our city in history. So, What about the driver-less cars? 
There would be many imaginations about future, but machine learning could help us to build a city with a new struture. 
The point is to create new structure of city. 
Local layer would be more like groups with less road going through.
Most of the needs can be met within the group
District Layer is more pedestrian friendly, allowing longer walking routes.
It also includes network, for supply or fast driver-less.
Urban Layer focuses on urban centers, such as public service buildings or transportation hubs.
Public centers serve blocks, while transportation hubs serve a larger area.
Regional Layer is about connection. The whole city connected by a fast-travel network with diver-less cars.
And some bigger knots are connected by high-speed railway to another city.
Driver-less could hand with complex condition. So we can define knots for different layers, which connected by shortest path.
A smart car could choose a fastest way from one knot to another.
Different structure of layers would means different size of area. 
Overlaid fields together to get a new city structure. 
Mixed satellite maps with structure.
[Output of Machine Learning]
Different kind of space in the city. 
It would be easy to figure out the properties of each part, but would be hard to define functions of certain area.
[Details]
Buildings and outdoor space can be distinguished. But there will be more ambiguity in between-space. It is not so easy to define a park or a plaza.
Nolli Map 
Activity Space 
In a close distance, more type of space could be recognized.
[Function]
Some kinds of space could be defined. Although it never be the only answer.
A{AI} 