vizdoom/navigation
Visual navigation
Navigate a scenario from pixels under a bounded action budget.
Upstream publisher Farama FoundationFarama-Foundation/ViZDoom
Programmatic Doom scenarios for visual navigation, perception, planning, and combat.
WHAT THIS PROJECT MEASURES
VizDoom offers a practical, recognizable visual environment with programmatic state and actions. Real-time play makes inference latency part of the result.
Trunchbull should publish real-time and paused or turn-normalized modes as distinct releases so their scores are never presented as directly comparable.
EXECUTION CONTRACT
Requires a visual game sandbox, deterministic scenarios, action timing, recordings, and separate timing-mode releases.
TRUNCHBULL AVAILABILITY
Not yet imported
Not yet imported. Define distinct real-time and paused evaluation contracts before publishing results.
2 CASES
vizdoom/navigation
Navigate a scenario from pixels under a bounded action budget.
vizdoom/combat
Balance perception, movement, and action timing in a scored scenario.
2 REQUESTED TOOLS
Returns the current rendered frame.
Required
Applies movement and interaction inputs.
Required
MACHINE-READABLE PROVENANCE
{
"name": "VizDoom",
"catalogOwner": null,
"source": {
"url": "https://github.com/Farama-Foundation/ViZDoom",
"version": "Upstream project",
"commit": "Upstream head not pinned",
"license": "MIT"
},
"port": null,
"release": {
"digest": "Not compiled",
"lane": "sandbox",
"sandboxRequired": true
},
"cases": [
"vizdoom/navigation",
"vizdoom/combat"
],
"tools": [
"screen capture",
"game actions"
]
}BUILD ON THIS PROJECT