Perception
Read frames, identify scene changes, and keep uncertain observations separate from facts.
The agent needs perception, planning, control and memory to explore a city without a fixed script. This page describes the intended modules and their publication state.
← Back to capabilities overviewAll six are design targets. No module is presented here as a verified GTA VI integration.
Read frames, identify scene changes, and keep uncertain observations separate from facts.
Track position, heading and reachable space as the camera and traffic change.
Choose a destination, travel toward it, and revise the route when it fails.
Issue timed driving inputs and measure whether the vehicle moved as intended.
Detect a stuck state, collision or failed interaction and try a better action.
Keep useful experience across sessions and measure whether it improves future decisions.
When public runs begin, each capability should link to an event, result and relevant media.
There is no public GTA VI execution trace attached to these modules yet.