RL-based Control of a Lunar Lander

Lunar Landing Algorithm Comparison

We compare the performance of four different algorithms for controlling lunar lander simulations:

DQN
Q-Learning
SARSA
Monte Carlo
Random

For each algorithm, we provide both a simulation GIF and a trajectory plot illustrating the landing path.

DQN

DQN Lunar Landing Simulation

DQN Trajectory Plot

DQN: Simulation (left) and trajectory plot (right).

Q-Learning

Q-Learning Lunar Landing Simulation

Q-Learning Trajectory Plot

Q-Learning: Simulation (left) and trajectory plot (right).

SARSA

SARSA Lunar Landing Simulation

SARSA Trajectory Plot

SARSA: Simulation (left) and trajectory plot (right).

Monte Carlo

Monte Carlo Lunar Landing Simulation

Monte Carlo Trajectory Plot

Monte Carlo: Simulation (left) and trajectory plot (right).

Random

Random Lunar Landing Simulation

Random Trajectory Plot

Random: Simulation (left) and trajectory plot (right).