scienceworld-object-locator

To locate a target object in ScienceWorld, the skill systematically teleports the agent to and examines rooms. It parses environmental observations from each room to identify the specified object.

The best way to search for items in ScienceWorld is to use prioritized room teleportation. This approach evaluates candidate rooms to determine the most efficient search order before examining them.

Yes, you can prioritize candidate rooms during environment exploration in ScienceWorld. The skill supports optional candidate room prioritization to guide the search algorithm more efficiently.

An AI agent identifies a target object by leveraging the 'look around' command to observe the environment. It then parses these environmental observations to find the specified item.

If your agent cannot find a required task item, use an object locator to iterate through all available rooms. It handles object identification through environmental observation parsing until the target is found.

Yes, the object locator supports task completion for game AI by streamlining environment exploration. It locates necessary objects like a thermometer or metal fork to help agents finish their tasks.