Interaction Dynamic

Interaction dynamics research focuses on understanding and modeling how agents, whether robots, humans, or LLMs, interact and influence each other's behavior. Current research emphasizes learning these dynamics from diverse data sources, including human-human and human-robot interactions, using models like Hidden Markov Models, Variational Autoencoders, and Graph Neural Networks to capture complex, multimodal information and predict future interactions. This work is crucial for advancing human-robot collaboration, improving AI safety (especially in multi-agent systems), and creating more natural and effective human-computer interfaces.