AI for Situated Agents and Robots

Description

Current conversational agents (those with which you can interact with voice on your smartphones, or else chatbots), are limited in the applications they can support because they are unable to understand the situation and the context. This research intends to develop the technology (audiovisual scene understanding and deep learning) necessary for a machine to understand the environment and the sequence of activities and actions taken by a human so that it can intervene more effectively in these activities. We plan to develop a system that assists a person in carrying out a task, such as preparing a recipe (cooking), but studying other activities are welcome too.