Papers
7
Total Citations
113
H-Index
4
About
Haitao Meng is a leading researcher in real-time 3D perception for autonomous systems, focusing on stereo depth estimation and efficient 3D object detection. His core contributions lie in bridging the gap between high-accuracy deep neural networks and the stringent real-time, energy-efficient requirements of mobile robotics and autonomous driving. Meng pioneered the use of binary neural networks for stereo estimation, culminating in **StereoEngine** (45 citations), an FPGA-based accelerator that demonstrated real-time, high-quality depth estimation with drastically reduced storage and power consumption. He extended this work with GPU-accelerated implementations and the resource-efficient **Lite-Stereo** architecture, collectively establishing a new paradigm for deployable stereo vision. More recently, Meng has advanced 3D object detection without expensive LiDAR, developing efficient frameworks like **ER3D** and pseudo-LiDAR-based methods that maintain high accuracy while enabling real-time performance. His work on **OmniStereo** further pushes boundaries by enabling real-time omnidirectional depth sensing from fisheye cameras. With over 110 citations across his publications, Meng’s research is instrumental in making sophisticated 3D perception practical for resource-constrained platforms, directly impacting the deployment of autonomous vehicles and robots in the real world.
Research Focus
Key Achievements
Top Papers
- 1
- 2GPU-Accelerated Real-Time Stereo Estimation With Binary Neural Network38 citations · 2020
- 3Efficient 3D Object Detection Based on Pseudo-LiDAR Representation15 citations · 2023
- 4
- 5
- 6
- 7