上一篇文章講述了openpilot在ubuntu上的構建:https://www.cnblogs.com/torch-hxm/p/15143407.html

然而，並不知道這樣做下來有什麼用....（oxo)........

算了，一點點試吧，今天在瀏覽檔案的時候，看見官網的檔案裡說：

Testing on PC
------
For simplified development and experimentation, openpilot can be run in the CARLA driving simulator, which allows you to develop openpilot without a car. The whole setup should only take a few minutes.

Steps:
1) Start the CARLA server on first terminal
```
bash -c "$(curl https://raw.githubusercontent.com/commaai/openpilot/master/tools/sim/start_carla.sh)"
```
2) Start openpilot on second terminal
```
bash -c "$(curl https://raw.githubusercontent.com/commaai/openpilot/master/tools/sim/start_openpilot_docker.sh)"
```
3) Press 1 to engage openpilot

See the full [README](tools/sim/README.md)

好傢伙，這難道就是openpilot在PC端的測試方法？於是我找到了tools/sim/README.md，查看了裡面的內容，如下：

openpilot in simulator
=====================


## Running the simulator

First, start the CARLA server.
```
./start_carla.sh
```

Then start bridge and openpilot.
```
./start_openpilot_docker.sh
```

To engage openpilot press 1 a few times while focused on bridge.py to increase the cruise speed.

## Controls

You can control openpilot driving in the simulation with the following keys

|  key  |   functionality   |
| :---: | :---------------: |
|   1   |  Cruise up 5 mph  |
|   2   | Cruise down 5 mph |
|   3   |   Cruise cancel   |
|   q   |     Exit all      |

To see the options for changing the environment, such as the town, spawn point or precipitation, you can run `./start_openpilot_docker.sh --help`.
This will print the help output inside the docker container. You need to exit the docker container before running `./start_openpilot_docker.sh` again.

## System Requirements

openpilot doesn't have any extreme hardware requirements, however CARLA is very resource-intensive and may not run smoothly on your system. For this case, we have a low quality mode you can activate by running:
```
./start_openpilot_docker.sh --low_quality
```
NOTE: [CARLA requires](https://carla.readthedocs.io/en/latest/build_docker/) an NVIDIA graphics card.

You can also check out the [CARLA python documentation](https://carla.readthedocs.io/en/latest/python_api/) to find more parameters to tune that might increase performance on your system

## Further Reading
The following resources contain more details and troubleshooting tips.
* [CARLA on the openpilot wiki](https://github.com/commaai/openpilot/wiki/CARLA) 

進入資料夾，執行如下命令：

./start_carla.sh

然後他提示：

Nvidia docker is required. Re-run with INSTALL=1 to automatically install.

我嘗試在./start_carla.sh後面加上INSTALL=1可惜失敗了，所以，我直接改了start_carla.sh，改動後如下：

 1 #!/bin/bash
 2 
 3 #Requires nvidia docker - https://github.com/NVIDIA/nvidia-docker
 

 4 if ! $(apt list --installed | grep -q nvidia-container-toolkit); then
 5   #if [ -z "$INSTALL" ]; then
 6   #  echo "Nvidia docker is required. Re-run with INSTALL=1 to automatically install."
 7   #  exit 0
 8   #else
 9     distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
10     echo $distribution
11     curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add -
12     curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list
13     sudo apt-get update && sudo apt-get install -y nvidia-container-toolkit
14     sudo systemctl restart docker
15   #fi
16 fi
17 
18 docker pull carlasim/carla:0.9.11
19 
20 echo "#################pull finish##############"
21 
22 docker run \
23   --rm \
24   --net=host \
25   -e DISPLAY= \
26   -e SDL_VIDEODRIVER=offscreen \
27   -it \
28   --gpus all \
29   carlasim/carla:0.9.11 \
30   ./CarlaUE4.sh -opengl -nosound -quality-level=Epic
31   
32  echo "##################run finish##############"

然後再次執行這個.sh檔案，自動安裝所缺的元件，然後，又提示缺少docker，依照提示進行安裝。安裝之後，這個docker並不能直接使用，回報粗，還需要進行如下配置：

sudo chmod a+rw /var/run/docker.sock

完成後，輸入docker version，如果輸出如下就說明docker可以用了：

再次執行start_carla.sh，會下載很多東西，耐心等待。

未完待續.....

2021年8月16日14:36:17