The researchers’ approach, which they call AV-Map, aims to convert short videos with multichannel audio into 2D floor plans. A machine learning model leverages sequences of audio and visual data ...