OscarHuangWind / DRL-Transformer-SimtoReal-Navigation

[T-ITS'23] Sim-to-real goal-oriented mapless autonomous navigation (DRL navigation).
MIT License
88 stars 13 forks source link

Convergence curve comparison #19

Open asliyigit2 opened 3 weeks ago

asliyigit2 commented 3 weeks ago

Thank you for your work.

I reviewed your codes and trained them for 500 episodes.

44_grafik (copy)

However, when I examined the code, I could not find how you drew the Convergence curve comparison graph in Figure 5 in the article.

I share the chart below. I would like your guidance and help in drawing the graph given in Figure 5.

Figure5

OscarHuangWind commented 2 weeks ago

Thank you for your attention.

Figure 5 plots the mean and variance of the moving average reward, which means if you got five moving average rewards (the orange curve in your training reward plot) with five random seeds, you can obtain mean and variance as Fig. 5 shows.

asliyigit2 commented 2 weeks ago

Thank you for your returns.

Can you share the source codes?

OscarHuangWind commented 1 week ago

Sorry for late reply.

I just checked out my workspace and unfortunately, I could not find the source code of plotting I utilized for this work, since this is the one I developed around two years ago and has been already cleared from local PC after publishing the main code in Github.