UnrealCV Environment Interaction Script

Job ID: 40067868

Budget: $750 – $1,500 USD

I already have the overall concept defined: load any suitable 3D level (an UnrealZoo binary is fine, but your own map also works), drop in either a drone-style camera or a human first-person pawn, and then drive that pawn through the scene for 30 to 60 seconds while UnrealCV captures everything.

Here is exactly what I need you to hand back:

• A working Unreal Engine binary with Blueprints for the pawn or drone camera and any helper components needed for movement and recording (can load from UnrealZoo or Unreal Engine Fab Marketplace).
• A Python script that talks to UnrealCV. The script must be able to switch between random wandering and a more deliberate, “intelligent” path so I can see both modes in action during one run. While the agent moves, UnrealCV should record the viewport to an .mp4 and also dump any control data you generate (keyboard, mouse, nav points, etc.) into a serialized file.
• A lightweight GitHub repo containing all source, along with a concise README that walks me through downloading the level, opening the project, installing dependencies, and running the Python script end-to-end.

Keep the toolchain tight—Unreal Engine, Blueprint, Python, UnrealCV, and standard video/serialization libraries are all that should be necessary.

When you apply, focus on the substance: give me a short, detailed outline of how you will set up the scene, orchestrate the two navigation modes, and manage the capture pipeline. I’m less interested in a résumé dump and more interested in seeing your proposed workflow.