carlagymenv

This is a Carla Simulator and Gym Carla Environment

Setup

Carla Simulator

Please go to Carla Simulator Website to install Carla For this repository, the version Carla 0.9.9.4 is used.

• Unpack the Download .tar.gz File
• chmod +x CarlaUE4.sh
• Get some try with the codes under /PythonAPI/examples
• /PythonAPI/codes/egovehicle_radius.py is the code, which should work for our need.

How to run Carla

Launch a terminal in this folder and execute the simulator by running

 ./CarlaUE4.sh

This will launch a window with a view over the city. This is the view, you can fly around the city using the mouse and WASD keys. This simulator acts now as a server, waiting for a client to connect and interact.

Try an example to interact with the server

 ./spawn_npc.py -80

This should launch 80 Vehicles into the simulator.

Try also manual control the vehicle by running

 ./manual_control.py

Now we can test '''egovehicle_radius.py''' with Town04 by running

 ./egovehicle_radius.sh 

Gym Environment

Please go to gym github website Website to clone openai gym Gym environment here clone to the folder /PythonAPI/codes/ inside the carla

git clone https://github.com/openai/gym 

Now go inside the gym folder under /gym/envs/, create a new folder carla for the carla gym environment For the Carla Environment, they should two files unter the folder carla, init.py and carla_env

the file /gym/envs/carla/init.py should have:

from gym.envs.carla.carla_env import CarlaEnv

And also /gym/envs/init.py should have:

register(
    id='carla-v0',
    entry_point='gym.envs.carla:CarlaEnv',
    kwargs = {'host':'127.0.0.1', 'port': 2000, 'tm_p': 8000, 'filtervehicle':'vehicle.*', 'coord_file': 'map04_coordination_1.csv',
              'map_name':'Town04', 'num_vehicle': 10, 'velocity': 80, 'autopilot': True},
)
register(
    id='carla-v1',
    entry_point='gym.envs.carla:CarlaEnv',
    kwargs = {'host':'127.0.0.1', 'port': 2000, 'tm_p': 8000, 'filtervehicle':'vehicle.*', 'coord_file': 'map04_coordination_1.csv',
              'map_name':'Town04', 'num_vehicle': 10, 'velocity': 90, 'autopilot': False},
)

Under /gym/ in terminal, by running:

pip install -e .

This should be run, each time if the environment is changed.

Python Packages

Create a Python Conda Environment, by using

conda create --name carlagym 
conda activate carlagym

install python packages with /PythonAPI/codes/requirements.txt:

pip install -r requirements.txt

Hierarchy

Under /codes/simulate/: egovehicle_radius.py: Basically, we transformed this file to the carla_env.py, it measured the 15 Cars surrounding around the ego vehicle.

tm_audi.py: Try out with Traffic Manager and ego vehicle, played with parameters.

view_spawn_points.py: It can save the coordination points of the vehicles in 4 lanes, and also to view the trajectory of the vehicles(visualization)

view_way_points.py: Just plot out the way points

Under /gym/gym/envs/carla carla_env.py: This is the gym Environment for carla, Collision Sensor haven't be fixed yet.