Chaokang Jiang

About

Chaokang Jiang is a researcher working at the intersection of computer vision, robotics, and 3D perception, with a focus on enabling machines to understand motion and depth in dynamic environments. His primary research areas include 3D scene flow estimation, visual odometry, and the use of generative adversarial networks (GANs) for unsupervised learning in 3D space. Jiang’s most notable contribution is the development of SFGAN, an unsupervised generative adversarial framework for learning 3D scene flow directly from the 3D scene itself. This work, cited 19 times, addresses the challenge of tracking the 3D motion of every point in adjacent point clouds—a critical capability for autonomous driving and service robots. By leveraging the continuity of motion in the macro world, SFGAN bypasses the need for costly labeled data. Jiang has also advanced visual odometry with his work on Pseudo-LiDAR for Visual Odometry (14 citations), which reimagines how continuous image inputs can be used for robust navigation and localization. His research is particularly impactful in fields requiring precise 3D motion perception, such as robotics and autonomous systems, where his methods offer practical, data-efficient solutions.

Research Focus

Key Achievements

Top Papers

1. 1
   SFGAN: Unsupervised Generative Adversarial Learning of 3D Scene Flow from the 3D Scene Self

   19 citations · 2021
2. 2
   Pseudo-LiDAR for Visual Odometry

   14 citations · 2023
3. 3
   SFGAN: Unsupervised Generative Adversarial Learning of 3D Scene Flow from the 3D Scene Self

   2 citations · 2021

Key Collaborators

Contact & Links