Yutian Wu

Papers

Total Citations

H-Index

About

Yutian Wu is a leading researcher in autonomous driving perception, specializing in 3D object detection from LiDAR point cloud data. Their work focuses on developing real-time, single-shot neural network architectures that balance computational efficiency with high detection accuracy—a critical challenge for deploying autonomous systems in dynamic environments. Wu's most impactful contribution is the Realtime Single-Shot Refinement Neural Network with Adaptive Receptive Field (2021), which has garnered 17 citations for its novel approach to dynamically adjusting feature extraction scales, significantly improving object localization in cluttered scenes. This work builds on their foundational 2020 study introducing the single-shot refinement framework, which demonstrated that end-to-end architectures could achieve competitive accuracy without multi-stage processing. Wu's research addresses the fundamental tension in autonomous driving between the need for rapid inference (real-time performance) and precise 3D spatial understanding, making their methods particularly valuable for embedded systems in vehicles and robots. By pioneering adaptive receptive field mechanisms for point cloud data, Wu has advanced the field's ability to detect small or partially occluded objects, directly contributing to safer autonomous navigation. Their work continues to influence subsequent developments in efficient 3D perception for intelligent transportation systems.