Boyao Han

Boyao Han

I am an undergraduate student (class of 2022) at the College of Computer Science and Electronic Engineering, Hunan University. I am currently a research assistant at the School of Data Science, The Chinese University of Hong Kong, Shenzhen, advised by Prof. Li Jiang. My research interests lie in Embodied AI and Computer Vision, with a focus on Vision-Language-Action / Video-Action Models, 3D Computer Vision, and multimodal learning.

GeoPredict: Leveraging Predictive Kinematics and 3D Gaussian Geometry for Precise VLA Manipulation

Published in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Highlight, 2026

Abstract

Vision-Language-Action (VLA) models achieve strong generalization in robotic manipulation but remain largely reactive and 2D-centric, making them unreliable in tasks that require precise 3D reasoning. We propose GeoPredict, a geometry-aware VLA framework that augments a continuous-action policy with predictive kinematic and geometric priors. GeoPredict introduces a trajectory-level module that encodes motion history and predicts multi-step 3D keypoint trajectories of robot arms, and a predictive 3D Gaussian geometry module that forecasts workspace geometry with track-guided refinement along future keypoint trajectories. These predictive modules serve exclusively as training-time supervision through depth-based rendering, while inference requires only lightweight additional query tokens without invoking any 3D decoding. Experiments on RoboCasa Human-50, LIBERO, and real-world manipulation tasks show that GeoPredict consistently outperforms strong VLA baselines, especially in geometry-intensive and spatially demanding scenarios.

BibTeX

@inproceedings{qian2026geopredict,
  title     = {GeoPredict: Leveraging Predictive Kinematics and 3D Gaussian Geometry for Precise VLA Manipulation},
  author    = {Qian, Jingjing and Han, Boyao and Shi, Chen and Xiao, Lei and Yang, Long and Shi, Shaoshuai and Jiang, Li},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year      = {2026},
  note      = {Highlight}
}

Recommended citation: Jingjing Qian, Boyao Han, Chen Shi, Lei Xiao, Long Yang, Shaoshuai Shi, Li Jiang. (2026). "GeoPredict: Leveraging Predictive Kinematics and 3D Gaussian Geometry for Precise VLA Manipulation." CVPR. (Highlight).
Download Paper