Final Project, part 2: City optimization

In the second part of the final project, you are going to implement an optimization algorithm to make your city a city that people would love to live in.

In total, your city consists of 1024 plots of land (32 x 32) and should be populated with the following buildings:

5% Skyscrapers
8% Highrises
25% Offices
37% Houses
15% Parks
the remaining plots can be whatever you want The houses and parks are already given, you don’t have to generate them.

Initializing the grid

Your first task will be to populate the grid so that the distribution above is respected, and you have a total of 1024 buildings. In the file city.py you will find a function called reset_grid(). In this function you have access to the grid that you must fill. The grid consists of 32 x 32 plots of land, each containing a building. And each building has a type. The following
types are available:

EMPTY
PARK
HOUSE
OFFICE
HIGHRISE
SKYSCRAPER

City has the following functions that you can use to place your buildings:

.clear()
Fills the city with empty plots.
.rows, .cols
Returns the number of plots in each direction of the grid (8 in this case).
.get_building(i, j)
Returns the building at position (i,j) in the grid. Where i is the row and j is the column.
.set_building(i, j, building)
Sets the building at position (i,j) to be of building instance.
.get_building_type(i, j)
Returns the building type at position (i,j) in the grid. Where i is the row and j is the column.
.get_plot_center(i, j)
Returns the center (in 3D) of the plot at position (i,j) in the grid. Where i is the row and j is the column.

Optimization

Your city is now populated, but it is probably not very pleasant to live in. As the sun rises and sets throughout the day the skyscrapers and high rises obscure many of the parks and houses from seeing the sunlight. Some houses might be very far from their office, or any recreational park. Skyscrapers might be right next to each other, causing one side to be completely dark.

In optimization.py there’s a function called step(). This function gets called every time you press the ‘Step’ button. You can also make the function run a number of times using a ‘Start’ and ‘Stop’ or in the background (without showing it on the screen) with ‘Optimize offline’. In this function you should iteratively improve the initial layout that you generated to comply with the rules you designed.

A naive way to do this is to simply generate another random layout of plots and see if it is better. Running this for a while will get you better and better layouts of your city, but it will take a long time.

There are many better ways to optimize the city in a more direct manner. You can learn about this in the following document. Try to come up with your own clever solution for improving this naive optimization. In your demo presentation you will show the iterative optimization process in real-time.

Of course, you need some way to define what a better city layout is. In the following example we try to raise the price value of all houses as much as possible. The price of a house increases according to the following criteria:

Houses should be as close as possible to offices
Houses should be as close as possible to parks
Houses and parks should receive as much light as possible during a day
No two skyscrapers or high rises can be next to each other (1 plot in between diagonally, horizontally and vertically).

You should come up with your own set of rules for your city. Implement this in the score() function in optimizer.py. You have access to function that computes the contribution of light from all light positions for the whole city. Look at compute_sunlight_scores(), which return the light contribution over 11 light positions.

Presentation and Report

At the end of the project you (or your group) will submit a report explaining how you designed your rules and implemented the optimization. In the report you should describe what you have done in both the building generation and optimization parts of the final project, and how you have done it. What challenges did you face during the optimization and how did you overcome them?
There will also be a small demo session when each group can briefly present their solution and give a live demo of the optimization algorithm. Here, you don’t have to go over all details but just highlight the most important points and original ideas. Each presentations including the live demo should take between 5-7 minutes, please do not go over time.

Minimum requirements

What follows are the minimum requirements for a passing grade.

Building generation

At least your building should satisfy or out-do the following minimum requirements:

Building made out of simple shapes (rectangles, triangles)
Building is textured decently
Building includes simple decorations (windows, doors)
Buildings fit inside a grid cell.

You will get extra points for the following:

Using randomness to generate a variety of buildings
Non-trivial shapes included in your building
Properly textured non-trivial shapes
Interesting extra decorations (e.g. cornices, quoins, balconies, be creative!)

City optimization

At least your city optimization should satisfy the following minimum requirements:

Computation of score for set of rules to be optimized
Optimization implemented in a naive manner (e.g. random layout on every iteration or randomly swapping two cell types)