search

zw199502 / RL_navigation

deep reinforcement learning for autonomous navigation
26 stars 6 forks source link

About Testing #5

Open 05063112lcs opened 2 months ago

05063112lcs commented 2 months ago

Hello, thank you very much for your work. I have read your article and conducted tests. When testing in scenario 1, the success rate was very high. However, when testing word2 and word3, I found that the agent was spinning in place and looping continuously. Could you give me some advice? image

zw199502 commented 2 months ago

Hello, thank you very much for your work. I have read your article and conducted tests. When testing in scenario 1, the success rate was very high. However, when testing word2 and word3, I found that the agent was spinning in place and looping continuously. Could you give me some advice? image

I can't make a clear conclusion because I haven't maintained this project for several years. The phenomenon of rotation in place may be caused by unexplored situations. Worlds 2 and 3 are totally new environments which have not been explored before, so our pre-trained model may fail in some cases or behave unnaturally.