In this blog, I will shortly report about the ongoing development progress.
Developement Log
- Sorted all lines of codes based on their respective segments.
- NEAT functions, Display Check and File Management code are now separate.
- Sorted all NEAT functions categorically.
- Categorically = e.g. network functions are grouped separately from species functions.
- Renamed some NEAT functions for better consistency.
- Formatted the structure of the code so everything looks consistent.
Notes
Some code in the core script is related to the config script, which has not been touched yet. I might cover that next. Especially the ‚InitializeRun‘ function, which gets everything started, contains such code.
Some player data might be problematic to put onto the UI. Reconsideration if every piece of player data is even needed should be taken.
Bonus Content
I really like to put related videos into these development blogs. They contrast the technical advancements quite nicely.
Last time I linked a video in which TASBot completed the game with a 0% completion rate. This time, the video linked below shows four speedrunners racing the game competetively (The run starts at 17:40). I chose to put this video here for two reasons. First of all, this video displays the desired outcome of the project – the Reinforcement Learning bot might be able to perform like the runners in the video if enough training time is provided. I believe that with the right means, the AI could even outperform them. Second, the video is very entertaining; even for people who do not know the game.
