path-explore

Autonomous exploration is handled by selecting the farthest reachable position from the agent's spatial map and navigating there using pathfinding primitives, then updating the map after the episode. The agent intelligently targets unexplored territory to systematically uncover new areas without direct player input.

Frontier-directed pathfinding works by identifying the farthest reachable point on the agent's internal spatial map and attempting to reach it using navigation primitives, updating the map dynamically after exploration.

Yes, you can use it to map procedurally generated game worlds without player input by having the autonomous agent systematically navigate toward unexplored territory to uncover new areas and discover points of interest.

The best way to discover points of interest is to instruct an autonomous agent to perform a frontier-directed exploration episode, which systematically targets and navigates toward the farthest reachable unexplored positions.

No dependencies are required to run autonomous navigation and exploration scripts, as the Skill operates independently using its own internal spatial maps and navigation primitives to manage pathfinding.

Your agent might stop exploring new terrain if it cannot find a reachable frontier position in its internal spatial map, as the Skill relies on selecting the farthest reachable point to trigger each exploration episode.