PI: Nuno Vasconcelos
Institution: University of California, San Diego
Although vision and language methods have made significant progress, it remains challenging to integrate these heterogeneous methods for general
tasks. We are investigating Vision-language Navigation (VLN) task, in which a robot in an unseen environment is navigated through natural language.
We will build an effective method for VLN tasks using a virtual environment which provides photo-realistic graphics inside buildings. This project aims for both application and top-tier conference publication.