Liu Liu

Collective Skill Tree Search transforms LLMs into versatile agents capable of mastering complex tasks through a structured skill tree that enhances their adaptability and performance.

Achieving high-fidelity 3D scene reconstruction from monocular video, ManiSplat enables robots to interact with their environments in a more controllable and realistic manner.

Forget painstakingly aligning objects in 3D scene generation; 3D-Fixer uses fragmented geometry as a spatial anchor, boosting accuracy while keeping things efficient.

By structurally disentangling temporal joint planning from frame-level manipulation, StructBiHOI achieves superior long-horizon stability and motion realism in bimanual hand-object interaction generation.

Achieve robust MoE inference on analog hardware without retraining by intelligently offloading only the most noise-sensitive experts to digital compute.

Skip the 3D modalities: Spa3R shows that strong spatial reasoning can emerge directly from 2D vision alone by learning to predict feature fields from multi-view images.

Achieve state-of-the-art image fusion in just one minute of training, bridging the gap between slow deep learning methods and fast but less adaptable traditional techniques.

Ditch continuous pose regression for articulated objects: DICArt's discrete diffusion approach unlocks more reliable 6D pose estimation by respecting kinematic constraints and navigating the search space more effectively.