W A S D : Move Forward/Left/Back/Right
SPACE / SHIFT : Move Up / Down
↑ ↓ ← → : Look Around
Q E : Roll Camera
Touch:
1 Finger Drag: Look Around
2 Finger Pinch: Zoom (Move Forward)
2 Finger Drag: Pan (Move Sideways)
System Setup
Mu (Gravity) — mass ratio of the two bodies (Earth–Moon ≈ 0.01215). It sets where the planets and all five Lagrange points sit; changing it repositions the whole system.
AI Control
Model — pick a trained PPO agent. The list runs from 0.5M (early) to 4.7M (best) to 5M (final), so you can see how training quality changes (the final isn't the best!).
RUN AI (BEST OF 20) — runs the chosen agent 20 times from your Initial State and plays back the best (highest-reward) attempt. The AI fires thrusters to hold the L1 orbit.
Initial State
Preset — snap the satellite's start to a Lagrange point (L1–L5), or choose Custom to type your own.
X / Y / Z — starting position in rotating-frame units (Earth sits at ≈ −Mu, the Moon at ≈ 1−Mu).
VX / VY / VZ — starting velocity along each axis.
Simulation
Duration (Steps) — how many timesteps to simulate (each step = 0.01 time units).
RUN PHYSICS SIMULATION — pure gravity, no AI and no thrust. Shows the natural, uncontrolled path so you can compare it against the AI.
Playback
Time slider — scrub through the computed trajectory.
▶ / ⏸ — play or pause. ⏮ — jump back to the start.
Visuals
Trail — show/hide the path line. Stars — toggle the starfield. Auto-Play — start playback automatically after a run finishes.
Trajectory Files
Choose File — load a previously exported trajectory (.json). EXPORT TRAJECTORY — download the current run as JSON.