Skip to content

Latest commit

 

History

History
173 lines (114 loc) · 6.75 KB

Install.md

File metadata and controls

173 lines (114 loc) · 6.75 KB
Raw

Installation

Requirements

To setup CARLA refer to the official CARLA documentation. Ensure all environment variables and dependencies are correctly configured.

Minimum Setup

The necessary setup to use the LunaticAgent class and the AgentGameLoop.py module is as follows:

# Clone the repository
git clone --recurse-submodules https://github.com/Daraan/LunaticAI-Driver-for-CARLA-Simulator.git LunaticAI
cd LunaticAI

# Install the dependencies
pip install -r docs/requirements/requirements.txt

# Use these two commands if you forgot the --recurse-submodules in the first step
git submodule init 
git submodule update

Leaderboard Setup

If you want to use the LunaticChallenger additional requirements are need to be installed to setup the Leaderboard 2.0. The scenario_runner is already included as a submodule.

Follow the instructions in the Leaderboard 2.0, however apply the following changes:

  1. Use the newer master branch instead of the leaderboard-2.0 branch:

    git clone -b master --single-branch https://github.com/carla-simulator/leaderboard.git

  2. If you are using Python 3.10+ skip the requirements installation; the leaderboard requirements are already included our in the .

  3. If you are using Python 3.9+ check if the pull request Python 3.9+ support{.github .fa .fa-github} is included yet; if not, apply the changes manually:

    # Alternatively to below code you can use the patch file from docs/requirements/leaderboard_8e5eda.patch

cd leaderboard
git remote add other https://github.com/Daraan/leaderboard-fork.git
git fetch patch upgrade-dependencies
git cherry-pick 8e5eda14e96c05387d9714086b5f7b90b13fd4df --no-commit

Documentation

If you locally want to build the documentation you need to additionally install Sphinx.

pip install -r docs/requirements/_readthedocs.txt
sh build_docs.sh

Clone this repository

# Navigate to a parent folder in which you want to store the repository
git clone --recurse-submodules https://github.com/Daraan/LunaticAI-Driver-for-CARLA-Simulator.git LunaticAI
cd LunaticAI
git submodule init 
git submodule update

Path Variables

Note

The following setup is only needed when you do not use a distributed version of the carla package or installed it via conda/pip, i.e. when import carla can be used without problems. Otherwise you need to either add the path to the carla .egg files to your PYTHONPATH environment variable or use the sys.path.append method in your script before importing carla.

Carla Folder

Setting the CARLA_ROOT variable allows you to import the carla module by one of the two workarounds below. It is recommended to still follow the below section for a more stable setup afterwards.

Temporarily set the path variables with the following commands or register them permanently in your system, e.g. via conda env var create. For Linux you can also modify your .bashrc or .bash_profile and for Windows you can use the setx command to set them permanently.

from launch_tools import carla

# or in this order
import launch_tools
import carla

Windows

set CARLA_ROOT=<path to carla folder>

Linux

export CARLA_ROOT=<path to carla folder>

Setting PYTHONPATH

If you do not use a packaged version of CARLA, e.g. installed via pip, but the distributed .egg files that come with carla locate the appropriate files and add them to your PYTHONPATH variable. You find them in ${CARLA_ROOT}/PythonAPI/carla/dist.

Adjust the file name depending on your system and Python version, for example it could look like this:

# Linux; temporary
export PYTHONPATH=$CARLA_ROOT/PythonAPI/carla/dist/carla-0.9.15-py3.10-linux-x86_64.egg:$PYTHONPATH

# or conda
conda activate <your_env>

conda env var create PYTHONPATH=$CARLA_ROOT/PythonAPI/carla/dist/carla-0.9.15-py3.10-linux-x86_64.egg:$PYTHONPATH

conda activate <your_env>

Important

Depending on your operating system, python and CARLA version the file will be named differently.

Troubleshooting

Importing Carla fails

There are multiple reasons for that. First consult the CARLA documentation and check if you have installed all necessary dependencies. The CARLA F.A.Q section provides further assistance..

AttributeError: `xml.etree.ElementTree.Element' object has no attribute 'getchildren'

This is caused by the fact that the xml.etree.ElementTree.Element.getchildren() method was removed in Python 3.9. Recent versions of the scenario_runner already include this patch. For the Leaderboard Setup see how to acquire this patch.

For all other cases, you can apply the patch manually at the error location; exchange:

- for elem in scenario.getchildren():
+ for elem in list(scenario):

ImportError: cannot import '...' from 'agents'

This indicates that another agents module is in your PYTHONPATH and is imported instead of the one from this repository. Make sure that the path to the cloned repository is in your PYTHONPATH before any other path. Modules that interfere are likely PythonAPI/carla/agents but also the leaderboard or scenario_runner might try to import the agents module before the one from this repository.

# A quick hack is to add the path to the repository at the beginning of your script
import sys
sys.path.insert(0, 'path to this project')

RuntimeError: Spawn failed because of collision at spawn position

An actor is spawned at a blocked position. Restart your simulation and try again.

'GLIBCXX_3.4.30' not found

In case you run into this error, be sure to first import carla and pygame afterwards.

ImportError: ./lib/libstdc++.so.6: version 'GLIBCXX_3.4.30' not found (required by /home/.cache/Python-Eggs/carla-0.9.15-py3.10-linux-x86_64.egg-tmp/carla/libcarla.cpython-310-x86_64-linux-gnu.so)

This issue is tracked here.

Rule Troubleshooting

For rules there is a separate project:./Rules.md#troubleshooting section.