search

TempleRAIL / nav-competition-icra2022-drl-vo

DRL-VO navigation policy for BARN Challenge
https://doi.org/10.1109/TRO.2023.3257549
MIT License
23 stars 2 forks source link

BARN Challenge Set Dynamic Obstacles #2

Closed nfhe closed 5 months ago

nfhe commented 5 months ago

I'm delighted to see that the authors have open-sourced such excellent navigation algorithms. However, I'm currently facing some challenges and hope you can assist me. I can run Ubuntu 20.04 on Docker with the robot navigation training you provided. After training the model, it performs well in the static BARN environment. Now, I want to run it in the dynamic BARN environment, but I couldn't find any tutorials in your repository.

If you could provide guidance on this matter, I would greatly appreciate it.

zzuxzt commented 5 months ago

Hi, thanks for your interest in our work. In this current repository, it sets the pedestrian kinematic maps to 0 values and works for the static BARN challenge. Since I am not familar with the details of the dynamic BARN challenge, I think you need to detect the dynamic obstacles first.You can then incorporate the detected kinematic information of dynamic obstacles into our pedestrian kinematic maps. The details of our DRL-VO control policy can be found in our paper "DRL-VO: Learning to Navigate Through Crowded Dynamic Scenes Using Velocity Obstacles". Hope it helps you in some way.

nfhe commented 5 months ago

Thank you very much for your response to my question. Thank you for your patient response and guidance.