Dantong Niu

I am a second-year Ph.D. student advised by Prof. Trevor Darrell at UC Berkeley.

I develop vision-language models for robotics.

Email  /  Google Scholar  /  Github

Selected Publications

I am fortunate that my PhD research unfolds in an era where the explosion of groundbreaking LLMs and VLMs is revolutionizing robotics learning with unprecedented possibilities.

LLARVA: Vision-Action Instruction Tuning Enhances Robot Learning

We propose LLARVA, a model trained with a novel instruction tuning method that leverages structured prompts to unify a range of robotic configurations and introduces the concept of visual traces to further align the vision and action spaces.

We present U2Seg, a unified framework for Unsupervised Universal image Segmentation that consistently outperforms previous state-of-the-art methods.

Previous Work

Before starting my Ph.D. studies, I worked on general machine learning and vision topics such as object segmentation, adversarial attacks, and video understanding.