Zhaoqi Leng

Papers

Total Citations

130

H-Index

About

Zhaoqi Leng is a leading researcher in 3D perception for autonomous systems, with a primary focus on point cloud processing and 3D object detection. His most significant contribution is the development of SWFormer (Sparse Window Transformer), a pioneering architecture that adapts transformer models to efficiently process sparse 3D LiDAR data. This work, cited over 126 times, introduced a novel sparse window attention mechanism that dramatically improves both accuracy and computational efficiency in detecting objects from point clouds—a critical capability for self-driving vehicles. Leng also advanced the field through LidarNAS, which unified and automated the search for optimal neural network architectures tailored to 3D point cloud data. His research directly addresses the fundamental challenge of balancing detection performance with real-time processing constraints in autonomous driving. By bridging transformer architectures with sparse 3D data, Leng has helped establish new paradigms for how machines perceive and understand three-dimensional environments, making his work essential reading for researchers developing next-generation perception systems for robotics and autonomous vehicles.