search

mit-acl / cadrl_ros

ROS package for dynamic obstacle avoidance for ground robots trained with deep RL
573 stars 157 forks source link

About the training set. #2

Closed chenzhiwo closed 6 years ago

chenzhiwo commented 6 years ago

How can I obtain the initial training set that the paper say that is relased to the public?

mfe7 commented 6 years ago

Thanks for pointing this out. I have just uploaded the files and some instructions to a Dropbox folder: https://www.dropbox.com/sh/yu1spzhhj8c9akl/AAALo8yXSfQ1nxaU2KFRGWuTa?dl=0

I have also added this info to this repo's readme. Let me know if there are any questions.

chenzhiwo commented 6 years ago

Thanks!

chenzhiwo commented 6 years ago

Here is a little problem, the dataset_readme.txt said that states.shape ---> (100000, 26), however, for the regular dataset, states.shape should be (100000, 28).

chenzhiwo commented 6 years ago

Just a little confusing, thank you!

mfe7 commented 6 years ago

Take a look at the description of the state array for the two styles of observations. One has 26 elements and the other has 28

Good point, I’ll try to make that more clear On Fri, Jun 8, 2018 at 2:59 AM chengzhi chen notifications@github.com wrote:

Just a little confusing, thank you!

— You are receiving this because you modified the open/close state.

Reply to this email directly, view it on GitHub https://github.com/mfe7/cadrl_ros/issues/2#issuecomment-395667743, or mute the thread https://github.com/notifications/unsubscribe-auth/ADWgpfACgjXQdtec8vJ9278NVTPEdOlkks5t6iDFgaJpZM4Udrqf .