Learnr Idle Mining

Imagine if your idle game characters can actually learn! Learnr Idle Mining is a NFT game where your characters evolve to get better at their tasks.

For this test project, these characters navigate through grid worlds. They collect rewards along the way which the player can use. Players can later optimize how their agents are trained' strategies through reinforcement learning, continuously improving their performance.

The twist? Each NFT becomes a living entity. They get smarter with every training session Built on top of Lilypad's computating platform. This shows how gaming and NFTs can level up!

NFTs are built using ERC1155 because they can be minted in batches, and are cheaper to use than the ERC721s. Explored two training styles SARSA and Q-Learning which have similar outputs. Both algorithms could be run in a way such that the policies and Q Tables can be separated out. NFTs hold these key components of the model since they are the brains behind the agents. These will drive the NFTs in their desired environment.

I used Lilypad because they were able to provide decentralised computing and used as a trustless system to verify the amount of training an NFT has undergone. This would keep the NFTs from trained on other computers. I built the website on React