Relational Reasoning

Relational Reasoning

 

Research in relational reasoning and multi-agent interaction modeling have broad impact in many research fields such as visual reasoning, dynamics modeling, multi-agent systems, perception, scene understanding, and autonomous driving. In the mobility domain, effective understanding of the situation and accurate trajectory prediction of interactive agents play a significant role in downstream tasks such as motion planning and decision making. In the context of automated vehicle research, we use explicit relational structure recognition and prediction via latent interaction graphs among multiple heterogeneous, interactive agents to forecast behavior in driving scenes. Beyond the benefits to the research community, relational reasoning research has significant social impact such as helping understand interactive human behaviors, learning and simulating visual physical dynamics, enabling safe and high-quality human-robot / robot-robot interactions, improving visual and logical reasoning for artificial general intelligence, enhancing intelligent transportation systems and logistics systems.