search

Shuijing725 / CrowdNav_DSRNN

[ICRA 2021] Decentralized Structural-RNN for Robot Crowd Navigation with Deep Reinforcement Learning
https://sites.google.com/illinois.edu/crowdnav-dsrnn/home
MIT License
97 stars 23 forks source link

Algorithm Problem #16

Open ShelyH opened 2 years ago

ShelyH commented 2 years ago

Hi, sorry to bother you again! This experiment uses the PPO algorithm to update the policy, but this is an online policy. Can I use an offline policy, such as SAC, to replace the PPO algorithm in my experiment?

Shuijing725 commented 2 years ago

Yes, I think so. Please let me know what results you get!

ShelyH commented 2 years ago

Sorry, so far I haven't learned a successful navigation strategy. I wonder if it is the way the policy is updated. Could you give me some suggestions? Thanks!

Shuijing725 commented 2 years ago

Without knowing your implementation details, I apologize that I cannot give very useful advice. The issue might be a bug in your code, unsuitable hyperparameters, or something else.