Dengxin Dai
Papers
5
Total Citations
225
H-Index
4
About
Dengxin Dai is a leading researcher in autonomous driving and robot perception, with key contributions in LiDAR data processing, motion prediction, and vision-and-language navigation. His work on real-time LiDAR segmentation, notably the "Multi-Scale Interaction" framework (95 citations), enables efficient semantic understanding on embedded platforms—critical for self-driving cars and robots with limited computational resources. In motion forecasting, his "Motion Transformer" (74 citations) advances multimodal behavior prediction by combining global intention localization with local movement refinement, directly improving safety in autonomous systems. Dai also pioneered long-range vision-and-language navigation with "Talk2Nav" (48 citations), creating a large-scale dataset and a dual-attention spatial memory model that bridges natural language instructions and robotic wayfinding. His more recent work on map-based depth priors and one-shot domain adaptation for semantic segmentation further demonstrates his commitment to practical, cost-effective solutions for real-world deployment. With over 200 total citations and a consistent focus on bridging simulation and reality, Dai’s research stands at the forefront of making autonomous vehicles and robots more perceptive, efficient, and human-friendly.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4Improving Depth Estimation Using Map-Based Depth Priors5 citations · 2022
- 5