Papers
7
Total Citations
52
H-Index
5
About
Haobo Xu is a leading researcher at the intersection of robotics, computer vision, and energy-efficient computing systems. His work focuses on enabling real-time, high-accuracy perception for autonomous machines, particularly through the acceleration of stereo vision and 3D point cloud processing. Xu’s most influential contribution is the **Dadu-Eye** stereo vision accelerator, which achieves an impressive 5.3 TOPS/W while processing 30 fps at 1080p resolution—a breakthrough that balances deep neural network accuracy with hardware efficiency for deployment on robots and drones. This work has garnered 19 citations and is complemented by his broader efforts to accelerate DNN-based 3D processing for mobile platforms. Xu also explores the fusion of disparate sensory data and the construction of digital twin-driven control platforms for industrial robots, bridging virtual and physical systems. His tutorial on efficient computing for robotics from a circuit and system perspective provides a foundational guide for designing low-power, high-performance robotic hardware. With a career spanning from foundational sensor fusion theory to cutting-edge neural processing unit (NPU) accelerators, Xu’s research is pivotal for the next generation of autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1Dadu-Eye: A 5.3 TOPS/W, 30 fps/1080p High Accuracy Stereo Vision Accelerator19 citations · 2021
- 2Accelerating DNN-based 3D point cloud processing for mobile computing11 citations · 2019
- 3On the Fusion of Disparate Sensory Data6 citations · 1992
- 4
- 5Perception planning in mobile robot navigation5 citations · 2002
- 6Dadu-SV: Accelerate Stereo Vision Processing on NPU4 citations · 2022
- 7Toward Efficient Computing for Robotics: From a Circuit and System View2 citations · 2022