An agent to solve the archery minigame in Nobody Saves the World by Drinkbox Studios.
I plan to implement agents with following various techniques just for fun.
- Rule-based, with a priori knowledge on when/how targets move
- Rule-based, but doesn't know the accurate information about targets' speed and trail
- Reinforcement Learning, with position as states
- Reinforcement Learning, but only image/graphic input
The unpacking scripts are in main.py
, and its usage as follows:
python main.py extract -i levels.dat
python main.py extract -i misc.dat
python main.py extract -i resources.dat
NStW uses the GameMonkey Script for scripting. Here's a tutorial on how to modding.
- Firstly, generate the patched version of the scripts.
compilegm patch.gm patch.gmb
python main.py patch levels.gmb patch.gmb patch_.gmb
- Then, inject the
patch_.gmb
file into game.
withdll -d:worker.dll NtSW.exe