Optimal Crowds 🔮

The optimal_crowds package simulates the dynamics of a crowd of human beings moving through an environment towards a target. Once the room is set up, an HJB equation analogous to that found in Bonnemain et al. is solved to find the optimal trajectories leading to the targets. The motion of pedestrians is then simulated using a Agent-Based Model where agents strive the follow the optimally chosen trajectories and avoid obstacles and others via an adaptation of the Generalized Centrifugal-Force Model (GCFM) detailed in Chraibi et al..

INSTALL 💻

To correctly set up your package:

• Have the python modules numpy, matplotlib, json and scipy installed
• Clone the repository inside a directory, you can do that by using the command 'git clone https://github.com/matteobutano/optimal_crowds' in you favorite bash terminal
• Inside the directory where you cloned the repo, not inside the repo itself, create a folder named 'rooms', where you will place your room configurations

CREATE YOUR ROOM 🔨

In the 'rooms' folder, you will place the .json files containing all the information about the configurations you wish to simulate. Place the room_test.json file into your 'rooms' folder and modify it. The main elements of the room configuration file are:

• room_legth and room_height: the extension along the x and y axes of the simulation room
• initial_boxes: telling the rectagular regions of the simulation room we initialize the agents. Each box is an array telling in order: the rectangle's center's x coordinate, y coordinate, horizontal width, vertical width, the value of the average density in the initial box, measured in ped/m² and all the targets of agents spawned in that box. Agents will reach one of the listed target depending on the strategy given by the HJB solution
• targets: the target areas of pedestrians' motion. Each door is an array telling in order: the door's center's x coordinate, y coordinate, horizontal width, vertical width
• walls: rectangular obstacles placed in the simulation room. Each wall is an array telling in order: the wall's center's x coordinate, y coordinate, horizontal width, vertical width
• holes: openings in walls. Each hole is an array telling in order: the the hole's center's x coordinate, y coordinate, horizontal width, vertical width
• cylinders: cylindrical obstacles in the simulation room. Each cylinder is an array telling in order: the cylinder's center's x coordinate, y coordinate, the cylinder radius

Preset Rooms

Here you can find a selection of rooms already configured:

• exit_opposite.json
• metro_station.json
• slalom.json

START YOUR FIRST SIMULATION ▶️

In the directory where you cloned the 'optimal_control' repository, create a python script 'run.py' with instructions:

1. 'from optimal_crowds import simulations', to import the simulation module.
2. 'simu = simulations.simulation('room', T, recompute)', to create the simulation room , where: room, must be a string with the name without extension of the room's configuration file saved in you 'rooms' folder; $T$ must be a float determining the max time in seconds you allow agents to exit the simulation room. If $T$ is too small, agents won't move from their initial positions; in that case, try increasing $T$. On the other hand, if $T$ is too large, permormance could be impaired. Recompute is boolean, and active the periodic computation of the optimal strategy with frequency given in simulation step in the config file.
3. 'simu.run(verbose, draw)', to execute the simulation , where: draw must be boolean. If True the simulation room and the agents are plotted at each time step; verbose must be a boolean. If True the simulation time in seconds and the number of exited agents are printed at each time step.
4. 'simu.draw_history()' to finally, plot the actual trajectory each agent followed to exit the simulation room.

CONTRIBUTE 🏁

If you like what this does, feel free to improve upon code. Just follow these steps to contribute:

1. Fork it
2. Create your feature branch (git checkout -b my-new-feature)
3. Commit your changes (git commit -am 'Add some feature')
4. Push to the branch (git push origin my-new-feature)
5. Issue a pull request