Set up CARLA simulator and program

git clone https://github.com/lzqw/CARLA.git

cd CARLA

wget https://leaderboard-public-contents.s3.us-west-2.amazonaws.com/CARLA_Leaderboard_2.0.tar.xz

tar -xf CARLA_Leaderboard_2.0.tar.xz

conda create -n carlapy python=3.7

conda activate carlapy

pip3 install -r requirements.txt

Run simulator

cd {PATH TO CARLA simulator}

./CarlaUE4.sh -quality-level=Low -world-port=2000 #quality:Low or Epic

cd {PATH TO CODE}

python run_perception.py --host 127.0.0.1

cd /opt/CARLA_Leaderboard_20 #server

./CarlaUE4.sh -carla-rpc-port=2000 -RenderOffScreen -graphicsadaper=1 #server

cd {PATH TO CODE} #local

python run_perception.py --host {IP OF SERVER} #local

/CODE/agent/test_agent/agent_config.txt 中设置手动或自动
在/CODE/agent/test_agent/agent.py中修改程序，run_step函数可获取周围物体的bounding box信息，并最终给出控制量。