search

hanruihua / rl_rvo_nav

The source code of the [RA-L] paper "Reinforcement Learned Distributed Multi-Robot Navigation with Reciprocal Velocity Obstacle Shaped Rewards"
MIT License
174 stars 31 forks source link

reward change chart #11

Closed FADEAWAY114 closed 1 year ago

FADEAWAY114 commented 1 year ago

How do I get the final reward change chart?

hanruihua commented 1 year ago

Hi, the average reward of each robot is saved in the list: ep_ret_list_mean. (https://github.com/hanruihua/rl_rvo_nav/blob/main/rl_rvo_nav/policy_train/multi_ppo.py#L256), you can plot the mean at each epoch to see the reward change chart.

FADEAWAY114 commented 1 year ago

嗨，每个机器人的平均奖励保存在列表中：ep_ret_list_mean。（https://github.com/hanruihua/rl_rvo_nav/blob/main/rl_rvo_nav/policy_train/multi_ppo.py#L256），您可以绘制每个时期的平均值以查看奖励变化图。

Thank you. Please ask which file is required for the example with moving obstacles？

hanruihua commented 1 year ago

You can change world name to be dynamic_obs_test.yaml to configure the environment with dynamic obstacles. However, the observations may be modified. I recommend to consider other robots as moving obstacles which is similar and easy to use.

FADEAWAY114 commented 1 year ago

Ok thanks. I'm going to close this issue.