Papers
20
Total Citations
205
H-Index
8
About
Dmitry Yudin is a robotics and artificial intelligence researcher whose work spans autonomous navigation, 3D scene understanding, and deep learning for mobile robotic systems. His research addresses fundamental challenges in enabling robots to perceive, localize, and navigate complex real-world environments with computational efficiency suitable for deployment on physical hardware. Yudin's most influential contribution, "Real-Time Object Navigation with Deep Neural Networks and Hierarchical Reinforcement Learning" (2020, 51 citations), tackled the critical gap between laboratory deep learning methods and practical robot deployment. His work on LiDAR-based localization and SLAM (38 citations) advanced reliable outdoor robotic navigation independent of lighting conditions. More recently, he has pushed into sophisticated 3D understanding with domain-adaptive LiDAR point cloud segmentation and open-vocabulary object grounding using 3D scene graphs — enabling robots to interpret ambiguous natural language descriptions within spatial contexts. Notable among his recent contributions is PRISM-TopoMap, addressing scalable long-duration mapping, and RozumFormer, exploring multimodal transformer architectures for language-conditioned robot manipulation. With over 160 cumulative citations across a decade of work ranging from early landmark-based navigation to cutting-edge vision-language models, Yudin represents a researcher consistently bridging theoretical advances in AI with the demanding constraints of real robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2Real-Time Lidar-based Localization of Mobile Ground Robot38 citations · 2021
- 3DAPS3D: Domain Adaptive Projective Segmentation of 3D LiDAR Point Clouds13 citations · 2023
- 4Beyond Bare Queries: Open-Vocabulary Object Grounding with 3D Scene Graph11 citations · 2025
- 5HPointLoc: Point-Based Indoor Place Recognition Using Synthetic RGB-D Images11 citations · 2023
- 6
- 7
- 8
- 9
- 10