phuongboi / lunar-lander

Solve lunar lander problem with reinforcement learning

This repo contain some of my re-implement of lunar lander with reinforcement learning algorithm

Lunar lander v1

• Solve lunar lander problem from openai Gymnasium use Q-learning and experience replay memory [2].
• The implementation base on fakemonk1 [1] and references from juliankappler [3]
• Friendly and simple implementation with pytorch
• Run python lunar_lander_v1.py

Training result

Lunar lander v2

• Solve lunar lander problem from openai Gymnasium [2] use Q-learning.
• Periodly update q_target network parameter [4]
• Use softmax policy instead of epsilon greedy policy
• Multiple training step from replay memory

Reference

• [1] https://github.com/fakemonk1/Reinforcement-Learning-Lunar_Lander
• [2] Mnih, Volodymyr, et al. "Playing atari with deep reinforcement learning." arXiv preprint arXiv:1312.5602 (2013).
• [3] https://github.com/juliankappler/lunar-lander