[vishwagw]

Uploading through the GitHub website. Make branch fix/camera-and-wind-examples with the first upload, then upload the rest to the same branch, one folder at a time, with these commit messages. Each upload replaces the existing file.

1. examples/core/04_camera.py:
   examples: fix camera example for camera_angle_degrees
2. examples/core/10_custom_wind.py:
   examples: fix custom wind example for np.random.Generator
3. PyFlyt/core/abstractions/base_wind_field.py:
   core: use standard_normal in the WindFieldClass docstring example
4. docs_source/documentation/core/wind.md:
   docs: document np_random as an np.random.Generator
5. docs_source/documentation/core/abstractions/base_drone.md:
   docs: document np_random as an np.random.Generator on the drone page

---

Description

Two of the core examples fail on the current code.

examples/core/04_camera.py (#71) still passes camera_angle, which QuadX renamed to camera_angle_degrees:

TypeError: QuadX.__init__() got an unexpected keyword argument 'camera_angle'

• It now passes camera_angle_degrees.
• The step-count comment now says 100 steps (~1 second), matching the loop. It previously said 1000.
• The example drew the RGBA, depth and segmentation images onto the same axes, so only the segmentation image was visible. They're now shown side by side.

examples/core/10_custom_wind.py calls self.np_random.randn(). The Aviary passes wind fields an np.random.Generator (np.random.default_rng()), which has no randn, so the example fails on its first step:

AttributeError: 'numpy.random._generator.Generator' object has no attribute 'randn'. Did you mean: 'random'?

• It now uses standard_normal(wind.shape).
• The same line is in the WindFieldClass docstring, which the wind docs page shows as the custom wind example, so that line is updated too.
• The wind and drone docs pages listed np_random as np.random.RandomState. They now say np.random.Generator, which matches the type hints in WindFieldClass and DroneClass.

Fixes #71

Testing

Run on Linux under xvfb with Python 3.12, numpy 1.26.4, pybullet 3.2.7 and matplotlib, with plt.show() saving the figure:

• I ran all ten examples in examples/core on master. 04_camera.py and 10_custom_wind.py failed with the errors above, and the other eight ran to completion. With this PR, both run to completion: 10_custom_wind.py runs all 1000 steps.
• The camera example's figure now shows all three images (attached).
• Pre-commit passes on the changed files, apart from pyright, which reports gymnasium 1.4 spaces.Dict typing errors in other files on master too.