# CARLA Setup
Follow the steps in [Running in a Docker - CARLA Simulator](https://carla.readthedocs.io/en/latest/build_docker/).
1. [Install docker](#Install-docker)
2. [Install nvidia-docker](#Install-nvidia-docker)
3. [Running CARLA RL](#Running-CARLA-RL)
:::info
Add `sudo` before any `apt` or `docker` comment, or if permission denied message received.
:::
## Install docker
Refer to: https://docs.docker.com/engine/install/ubuntu/
```
$ sudo apt-get update
$ sudo apt-get install \
apt-transport-https \
ca-certificates \
curl \
gnupg-agent \
software-properties-common
```
If there are any package not updgraded, run:
```
$ sudo autoremove
$ sudo apt upgrade
```
Next,
```
$ curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo apt-key add -
$ sudo apt-key fingerprint 0EBFCD88
$ sudo add-apt-repository "deb [arch=amd64] https://download.docker.com/linux/ubuntu \
$(lsb_release -cs) \
stable"
$ sudo apt-get update
$ sudo apt-get install docker-ce docker-ce-cli containerd.io
```
Run a test
```
$ sudo docker run hello-world
```
## Install nvidia-docker
Refer to: https://github.com/NVIDIA/nvidia-docker
```
$ distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
$ curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add -
$ curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list
$ sudo apt-get update && sudo apt-get install -y nvidia-container-toolkit
$ sudo systemctl restart docker
```
Test nvidia-smi with the latest official CUDA image
```
$ sudo docker run --gpus all nvidia/cuda:10.0-base nvidia-smi
```
:::info
In new version, use `docker run` instead of `nvidia-docker run`.
Remember to add `sudo`.
:::
## Running CARLA RL
Refer to:
https://carla.readthedocs.io/en/latest/build_docker/
https://github.com/carla-rl-gym/carla-rl
```
$ sudo docker pull carlasim/carla:0.8.2
```
Clone the carla-rl github repo
```
$ git clone https://github.com/foojiayin/carla-rl.git
$ cd carla-rl
```
:::info
My fork: https://github.com/foojiayin/carla-rl.git (Added some unstated dependencies)
Original repo: https://github.com/carla-rl-gym/carla-rl.git (If you clone this repo, you might need to **[modify some files](#Modified-files)**
:::
### Building server
Build and run server in docker.
```
$ sudo docker build server -t carla-server
$ sudo docker run --rm -it --gpus all -p 2000-2002:2000-2002 carlasim/carla:0.8.2 /bin/bash
```
### Running server
Now we are in a container with username carla.
```
$ ./CarlaUE4.sh /Game/Maps/Town01 -carla-server -benchmark -fps=15 -ResX=800 -ResY=600 -carla-world-port=2000
```
- `Town01` can be replaced with other map
- port can be changed, too
#### Debug
-
:::danger
xdg-user-dir: not found
:::
***Solution***
Run carla version 0.8.2 instead of 0.9.9 (latest)
```
sudo docker run --rm -it --gpus all -p 2000-2002:2000-2002 carlasim/carla:0.8.2 /bin/bash
```
-
:::danger
docker: Error response from daemon: OCI runtime create failed
unsatisfied condition: cuda>=10.2, please update your driver to a newer version, or use an earlier cuda container
:::
***Solution***
Upgrade CUDA version： https://developer.nvidia.com/cuda-downloads
Choose OS version (BigWhite: Linux > x86-64 > Ubuntu > 16.04)
-
:::danger
segmentation fault SIGSEGV: invalid attempt to read memory at address 0x0000000000000010
:::
***Solution***
When [build and run the server](#Building-server): add `--gpus all`
```
$ ./CarlaUE4.sh /Game/Maps/Town01 -carla-server -benchmark -fps=15 -ResX=800 -ResY=600 -carla-world-port=2000
```
### Build and run client
- Before building client, please apply **[the following changes](#Modified-files)** if you clone the repo from https://github.com/carla-rl-gym/carla-rl
- Next, build and run the client.
```
$ sudo docker build client -t carla-client
$ sudo docker run -it --network=host --gpus all -v $PWD:/app carla-client /bin/bash
```
#### Debug
- During pygame installation
:::danger
Unable to run "sdl-config". Please make sure a development version of SDL is installed.
:::
***Solution***
In `DockerFile`, add
```
RUN sudo apt-get install -y libsdl-dev libsdl-image1.2-dev libsdl-mixer1.2-dev libsdl-ttf2.0-dev
RUN sudo apt-get install -y libsmpeg-dev libportmidi-dev libavformat-dev libswscale-dev
RUN sudo apt-get install -y python3-dev python3-numpy
```
### Training in the client
Now we are in a container with username `root`.
- Make sure the server is on. Next, run the client.
```
$ python client/train.py --config client/config/base.yaml
```
:::info
If the server port is not `2000`, please add `--starting-port=`, follow by the port number
:::
#### Debug
-
:::danger
TCP connection closed *(xN)*
:::
***Solution***
Please check that the server is running, and the `--starting-port` argument in client is consistent with the `--carla-world-port` in server.
- If error message (after client is successfully connected)
:::danger
assert dtype is not None, 'dtype must be explicitly provided. '
:::
***Solution***
Install gym version 0.10.8 instead of gym:
In `carla-rl/client/requirements.txt`: replace `gym` with `gym==0.10.8`
-
:::danger
no module named 'cv2'
:::
***Solution***
Install opencv.
In `carla-rl/client/requirements.txt`: add `opencv-python`
-
:::danger
from .cv2 import * ImportError: libSM.so.6: cannot open shared object file: No such file or directory
:::
***Solution***
In `carla-rl/client/DockerFile`: add
```
RUN sudo apt-get update
RUN sudo apt-get install -y libsm6 libxext6 libxrender-dev
```
## Modified files
- carla-rl/client/DockerFile
```
FROM anibali/pytorch
USER root
# Install vim for local development
RUN ["apt-get", "update"]
RUN ["apt-get", "install", "-y", "vim"]
RUN ["apt-get", "update"]
RUN ["apt-get", "install", "-y", "ffmpeg"]
# Add requirement.txt first for caching purposes.
COPY requirements.txt /app
RUN sudo apt-get install -y mercurial
RUN hg clone https://bitbucket.org/pygame/pygame
RUN cd pygame
RUN sudo apt-get install -y libsdl-dev libsdl-image1.2-dev libsdl-mixer1.2-dev libsdl-ttf2.0-dev
RUN sudo apt-get install -y libsmpeg-dev libportmidi-dev libavformat-dev libswscale-dev
RUN sudo apt-get install -y python3-dev python3-numpy
RUN sudo apt-get update
RUN sudo apt-get install -y libsm6 libxext6 libxrender-dev
RUN pip install -r requirements.txt
# Running a terminal lets you run any script.
CMD /bin/bash
```
- carla-rl/client/requirements.txt
```
Pillow
numpy
protobuf
pygame
matplotlib
future
tensorboardX
scikit-video
pyyaml
gym==0.10.8
IPython
opencv-python
```
- carla-rl/client/carla/client.py
```python=169
def _read_sensor_data(self):
while True:
data = self._stream_client.read()
if not data:
return
# raise StopIteration
yield self._parse_sensor_data(data)
```
#### Debug
:::danger
docker: Error response from daemon: Unknown runtime specified nvidia
:::
## Useful Commands in Docker
- Get container id or check status
```
sudo docker ps # Check current running containers
sudo docker ps -a # Check all containers
```
- Copy files into container (do this outside container)
```
sudo docker cp ${file-to-copy} ${container-id}:\${dest-location-in-docker}
## Example
# Copy file from computer to container
sudo docker cp copyme.txt c4d9f55f4db4:/home/carla/PythonClient
# Copy file from container to computer
sudo docker cp c4d9f55f4db4:/home/carla/CarlaUE4.sh ~/carla-rl/
```
- Do no use '~/' for home directory in container, use full directory, ex: ''/home/carla' (for server)
- The files in client '/app' refers to the computer '/carla-rl', do not need to use `docker cp`
- Open terminal in a running container instance
```
sudo docker exec -it ${container-id} bash
## Example
sudo docker exec -it c4d9f55f4db4 bash
```
## Output results
debug videos under `carla-rl/outputs/video` (show the last few minutes before collision / reach goal)
## Reproduce Benchmark
### A2C
To reproduce the results, run a CARLA server and inside the `carla-client` docker run,
`python client/train.py --config client/config/a2c.yaml`
### ACKTR
To reproduce the results, run a CARLA server and inside the `carla-client` docker run,
`python client/train.py --config client/config/acktr.yaml`
### PPO
To reproduce the results, run a CARLA server and inside the `carla-client` docker run,
`python client/train.py --config client/config/ppo.yaml`
### On-Policy HER
To reproduce the results, run a CARLA server and inside the `carla-client` docker run,
`python client/train.py --config client/config/her.yaml`
Sign in with Wallet
Connect another wallet