Today I kick-started this pet project of mine in an attempt to kill 2 birds with 1 stone.
1st bird: I have learned about the basic Genetic Algorithm but have not had the opportunity to apply it on real-world problems. Hence I thought this could help me to take it up a notch and gradually I would be able to apply it on greater problems.
2nd bird: Creating an app has always been one of my goals, however learning Android SDK, JS, HTML and Adobe PS or graphics in general seem a tad too daunting for me. It could take up to years too!
I am thinking of training an AI to play this game by following readily available tutorials.
Plan & Timeline:
- Build the app – 1 week
Took longer than expected due to some personal matters to attend to. Also I had been very busy at work from taking over my seniors work due to his departure from the company.
- Build AI using Genetic Algorithm – 1 week
Update 2016/12/31 – 11 pm
- Managed to get the world completed =D.
- Some notes regarding Box2D according to the tutor. The rest of the notes would be inline with the codes which I would share it in git upon completion.World
– – Mass
– – Velocity
– – Location
– – Angles
– – Fixtures : Physical attributes of bodies/ things inside the game world.
– – – – Shape
– – – – Density
– – – – Friction
– – – – Restitution
Update 2017/01/24 – 11 pm
- So I have managed to animate the monsters, enabled vision for Mario AI as input (the small box with squares), have music and sounds playing. The game was actually in full screen and that I have enabled auto scaling which is the reason why the hud, i.e.: score, time and level became very small. However, in fact it looks just fine from the actual app. It is just going to be screen shot for now as I am still rather occupied with work.
- Next, I am going to write Neural Network, trained using Genetic Algorithm ( to search for proper weight) to trigger 3 inputs which are Up, Left and Right.
Update 2017/07/11 – 6.13 pm
Time just flown by and I have yet to have the chance to complete this. I have made a lot of mistakes and I wish I knew better.