Dong Tian

Papers

Total Citations

H-Index

About

Dong Tian’s research lies at the intersection of 3D scene understanding and intelligent robotic navigation, with a focus on estimating motion from point clouds and improving path planning efficiency. His most notable contribution is the development of FESTA (Flow Estimation via Spatial-Temporal Attention for Scene Point Clouds), a novel framework that leverages attention mechanisms to accurately estimate scene flow—the 3D motion of points across time—from irregular point cloud data. This work, which has garnered over 43 citations, is critical for applications like autonomous driving, robot navigation, and augmented reality, where understanding dynamic 3D environments is essential. In addition, Tian has advanced mobile robot path planning by proposing a deep neural network approach to learn heuristic functions that guide search algorithms like A* and D*. This method significantly improves computational efficiency by better approximating true path costs, a key challenge in real-time navigation. With a growing citation record and contributions that bridge deep learning and robotics, Dong Tian is shaping how machines perceive and move through complex, dynamic 3D worlds.