单线程cpu1小时收敛的赛车自动驾驶训练

https://github.com/zdx3578/DDPG-Keras-Torcs

安装简介：

cd /home/wz/showcode

conda create -n DDPG_TORCS python=2.7                  #在anaconda下给python2.7建立一个新的环境

source activate DDPG_TORCS                                       #激活该环境

pip install keras==1.1.0                                                #在该环境下安装keras 1.1.0

pip install tensorflow==0.12.0rc0                                #在该环境下安装tensorflow 0.12.0rc0 

sudo apt-get install git

git clone https://github.com/yanpanlau/DDPG-Keras-Torcs.git

git clone https://github.com/ugo-nama-kun/gym_torcs.git

conda create -n GYM_TORCS                                       #在anaconda下给python3建立一个新的环境

source activate GYM_TORCS                                        #激活该环境

sudo apt-get install xautomation                                #安装xautomation (http://linux.die.net/man/7/xautomation) 

git clone https://github.com/openai/gym.git                #配置gym环境
cd gym
pip install -e .

cd ..

cd gym_torcs-master/vtorcs-RL-color  

sudo apt-get install libglib2.0-dev  libgl1-mesa-dev libglu1-mesa-dev  freeglut3-dev  libplib-dev  libopenal-dev libalut-dev libxi-dev libxmu-dev libxrender-dev  libxrandr-dev libpng12-dev

 ./configure
make
sudo make install
make datainstall

cd /home/wz/showcode/DDPG-Keras-Torcs    #在指定环境中把ddpg跑起来
source activate DDPG_TORCS  

pip install gym
python ddpg.py

。。。

掌握好了来面试哦！