Sim tryouts and selection

Goal: implement a basic tracked robot in PyBullet

What I did:
- Read start of PyBullet quickstart quide
- Gave up on PyBullet because it was insanely slow despite being "lightweight"
- Made a Mujoco demo with claude code and basic examples from repo
- read some mujoco docs - https://mujoco.readthedocs.io/en/stable/python.html
- researched isaac sim min specs

What worked:
- got both Mujoco and pybullet running on my mac
-

What failed:
- PyBullet speed was subpar (3fps rendering), and documentation was tragic (literally in a google doc)
- Mujoco was a little bit more overwhelming than i anticipated, but im catching on quick

Key learning:
- The documentation for these simulators is not the greatest
- PyBullet too slow + bad docs → switching to Mujoco
- Mujoco has steeper learning curve but better performance
- Mujoco uses data + model
- data contains the state of the sim, and we update it between time steps
- model contains scene and robot information, and defines how the simulator will update the data at each timestep
- Isaac sim is cool and probably the best option, but incredibly heavyweight (min spec rtx 4080 and 16gb vram)

Next session:
- add a basic scene (plane + cube) to Mujuco
- add a basic tracked robot (find online? use ai? differential drive)
- control basic tracked robot using keyboard
- update data.ctrl to control tracks
- eventually add zima stl files and get a proper simulation going
- capture camera data within simulation

Time spent: 2 hours