Mario Kart DS

Mario Kart DS (EU) reverse-engineering work: ASM-hacking and other tools

Mario Kart DS AI + overlays with DeSmuME, KCL/NKM parsers, and PyTorch tools.

MKDS AI + overlays with DeSmuME, KCL/NKM course parsers and PyTorch tools

Experimental javascript engine reimplemention of Mario Kart DS.

Experimental JavaScript engine reimplementation of MKDS (reads original track/model formats); inactive since 2018; forks xezno/mkjs, saavor/mkjs

AI-assisted matching decompilation of Mario Kart DS (EUR) — byte-perfect, oracle-verified

AI-assisted matching decompilation of Mario Kart DS (EUR) — byte-perfect, oracle-verified; no GitHub releases; repo created 2026-07-23

Repository dedicated to a DRL program that plays Mario Kart DS by itself

Deep-RL program that plays MKDS (2021)

A custom Gymnasium environment for Mario Kart DS

Gymnasium environment registered as gym_mkds/MarioKartDS-v0 (user ROM)

RL with Mario Kart DS

PPO agent learning to drive in MKDS on PyNDS (NooDS) with random savestate resets

Learning to play Mario Kart DS using Proximal Policy Optimization

Learning to play MKDS with PPO

Public research log: Mario Kart DS function mappings, decompilation findings, and native Rust port progress.

Public research log: function mappings, structure maps, state machines and progress on a native Windows Rust port

A Dissambly Of What Is On The Cart

Raw disassembly dump of the cartridge (minimal)

Reinforcement learning agent that learns to drive in Mario Kart DS using CNN+DQN. Used DeSmuME emulator.

CNN+DQN agent on DeSmuME

Reinforcement learning in Mario Kart DS using BizHawk emulator and pytorch

RL in MKDS using BizHawk + PyTorch

Reinforcement Learning with Mario Kart DS.

RL with MKDS via emulator Lua scripting