PPO algo performs badly

BarisYazici / deep-rl-grasping

Train deep reinforcement learning model for robotics grasping. Choose from different perception layers raw Depth, RGBD and autoencoder. Test the learned models in different scenes and object datasets

MIT License

192 stars 44 forks source link

PPO algo performs badly #29

Closed HarrisonC7 closed 3 months ago

HarrisonC7 commented 4 months ago

The PPO algorithm is difficult to converge，and the gripper always move up and away from the table.Could you please give me some hint about it.Sincerely appreciate it!

BarisYazici commented 4 months ago

I would recommend using SAC. It performed the best.

HarrisonC7 commented 4 months ago

I would recommend using SAC. It performed the best.

Thanks,SAC indeed did a wonderful job!