Week 8-9. Q learning - fixes

Since the last post I have focused more on solving some bugs and preparing the execution environment of the algorithm in cloud.

First of all, I have solved a bug that made the algorithm halve the iteration speed in a particular area of the map. This was due to my render configuration and my machine, but in cloud it is solved. On the other hand, I have gained access to a machine where a peak iteration speed of 18 is achieved. This is not a bad figure, however, I don’t expect much improvement either.

The good thing about this new environment is that the iteration speed will not be influenced, since it is a machine dedicated to this type of training. I hope that in the following weeks the development speed will increase thanks to this improvement.

On the other hand, and with less progress, I have been able to run some simulations in the cloud environment but without finding yet any better solution than the current one.