hccz95/decentralizedergodiccontrol

cloned from https://github.com/i-abr/ErgodicControl/

Decentralized Ergodic Control

Create a ros melodic environment with this package in the src folder. First source the ros devel/setup file To run use the following commands

python create_scene.py num_of_agents

where num_of_agents is replaced with the number of agents you want to see (dont set it high than 50).

then to visualize the environment run

rosrun rviz rviz

and open the default.rviz config in the rviz folder in this repo

Run in docker

1. Start a container by docker run -p 6080:80 hccz95/ros:noetic-dec. Or you can build from ./Dockerfile.
2. Open localhost:6080 in browser.
3. Open a terminal and compile the workspace:

   cd /home/catkin_ws/
   catkin_make
   echo "source /home/catkin_ws/devel/setup.bash" >> ~/.bashrc

4. Open a terminal (I) and run:

   roscore

5. Open a terminal (II) and run:

   cd /home/catkin_ws/src/DecentralizedErgodicControl/
   python create_scene.py 5

6. Open a terminal (III) and run rviz to render:

   cd /home/catkin_ws/src/DecentralizedErgodicControl/
   rosrun rviz rviz -d ./rviz/default.rviz

Run without ROS

地图大小为$[0,1]^2$

agent的state有4维，前两维是横纵坐标，后两维目前不确定，判断依据是obs_space中，前两维是$[0,1]$区间，后两维是$[-\inf,+\inf]$区间。

目前干了这么一些工作，把代码中涉及publish、rospy.Rate(1)、rospy.init_node调用的都删掉，暂时不再依赖roscore

但是发现一个问题，agent并没有按照预期分组，去覆盖不同目标；可能是因为去掉了rospy通信，导致agent各自为战，需要自己覆盖所有目标？

ergodic_controller.py里的self._ck_dict记录了ck的通信记录，用来算平均值的，可以从这里着手改进