Research Interest

The ultimate goal of my research is to develop trustworthy, interactive, and scalable models that integrate computer vision and natural language to enhance perception, understanding, reasoning, and decision-making in autonomous systems, such as autonomous vehicles, robots. To achieve this, my work leverages interdisciplinary approaches, unifying techniques from multimodal large language models, computer vision, robotics, machine learning, explainable AI, and reinforcement learning to advance autonomous systems capabilities. My prior work on Driving Behavior Analysis includes developing end-to-end deep learning and inverse reinforcement learning-based models for multimodal, context-aware, and interaction-aware motion forecasting. Additionally, I have extracted efficient driver embeddings from driving time series using scalable representation learning techniques and leveraged graph-based spatiotemporal convolutions to model interactive behavioral patterns among multiple agents within driving scenes.