Touchdown: Natural Language Navigation and Spatial Reasoning in Visual Street Environments
This work addresses the problem of natural language navigation and spatial reasoning for AI agents in real-life visual environments, representing a domain-specific advancement.
The authors tackled the problem of joint language-vision reasoning by introducing the Touchdown task and dataset, which requires an agent to follow navigation instructions and identify a location in a visual urban environment, with results showing that existing methods struggle on this dataset, as it contains 9,326 examples and presents an open challenge.
We study the problem of jointly reasoning about language and vision through a navigation and spatial reasoning task. We introduce the Touchdown task and dataset, where an agent must first follow navigation instructions in a real-life visual urban environment, and then identify a location described in natural language to find a hidden object at the goal position. The data contains 9,326 examples of English instructions and spatial descriptions paired with demonstrations. Empirical analysis shows the data presents an open challenge to existing methods, and qualitative linguistic analysis shows that the data displays richer use of spatial reasoning compared to related resources.