安装Mujoco、mujoco-py、multiagent-compete踩坑记录

安装Mujoco、mujoco-py、multiagent-compete踩坑记录

1、pip install 时报错“ValueError: Unable to determine SOCKS version from socks://127.0.0.1:1080/”

每次打开terminal都要执行 export all_proxy="socks5://127.0.0.1:1080"

2、Box 2D

从网上下载相应的安装包时要注意python的版本，cp37表示python3.7，另外还要看看文件名是不是正确，有的时候要更改文件名才能安装，可以查看pip._internal.pep425tags或者pip.wheel.pep425tags（这一步不是必须的）

3、报错“GLEW initalization error: Missing GL version”

要向环境变量中添加“export LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libGLEW.so”

sudo gedit ~/.bashrc
# 拉到最下面，复制粘贴：export LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libGLEW.so，ctrl+s 保存，关闭
source  ~/.bashrc

4、Failed to build mujoco-py

向环境变量中添加“export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/username(这里填写自己的用户名)/.mujoco/mujoco200/bin”方法同上

5、error: glew initalization error: missing gl version；GlfwError: Failed to initialize GLFW

安装glfw 安装完成后重启。

6、Failed to initialize GLFW；ERROR: could not initialize GLFW

先下载glfw，到官网上去下载source packge

export DISPLAY=:0 指定GPU

如果是通过ssh在服务器上使用mujoco，通常情况下是没法simulate的，也不能render！

7、ERROR: Could not build wheels for mujoco-py which use PEP 517 and cannot be installed directly

Same problem for me with the most recent release (mujoco 2.0.2.9). pip install mujoco_py==2.0.2.8 worked.

8、安装GLFW时：CMake Error at CMakeLists.txt:213 (message):RandR headers not found; install libxrandr development package

sudo apt-get install xorg-dev libglu1-mesa-dev

9、ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject

pip uninstall numpy
pip install numpy

10、PermissionError: [Errno 13] Permission denied: b’/home/whz/mujoco-py/mujoco_py/generated/mujocopy-buildlock’

sudo chmod 777 path-to-venv (这里要写自己环境的路径！) /lib/pythonX.X/site-packages/mujoco_py/generated/

11、Could not load dynamic library ‘libcudart.so.11.0’；dlerror: libcudart.so.11.0: cannot open shared object file: No such file or directory；ModuleNotFoundError: No module named ‘tensorflow.contrib’

TensorFlow版本太高了，换小于2.0版本的比如1.13.2

12、AttributeError: module ‘mujoco_py’ has no attribute ‘load_model_from_path’

gym版本太高了，换低一点的，比如0.9.1