This AI Mapping System Lets Robots See the World

Robots can now comprehend and carry out natural language commands with instance-level and attribute-level accuracy thanks to the robust framework for Vision-and-Language Navigation (VLN) introduced by the proposed Instance-aware Visual Language Map (IVLMap). In order to differentiate several instances of similar objects, IVLMap automatically combines RGBD visual data with a semantic map indexed by natural language in bird's-eye perspective.