Haoming Song
Papers
1
Total Citations
15
H-Index
1
About
Haoming Song is an emerging researcher at the forefront of embodied AI and robot learning, with a particular focus on the intersection of spatial reasoning, visual-language understanding, and robotic action planning. His most notable work, "SpatialVLA" (2025), has already garnered 15 citations within its first year — a strong indicator of its impact in a rapidly evolving field. In this paper, Song and his collaborators tackle a fundamental challenge in robot learning: how to equip Visual-Language-Action (VLA) models with rich spatial representations that generalize across diverse real-world environments. His innovative approach introduces pre-learned action grids that can be re-discretized to accommodate robot-specific spatial movements in novel setups, enabling both strong in-distribution generalization and impressive out-of-distribution adaptation. This contribution is particularly significant as it bridges the gap between high-level language-guided reasoning and low-level motor control — a critical bottleneck in deploying intelligent robots at scale. For students and researchers exploring robotics, foundation models, or human-robot interaction, Song's work represents a timely and technically rigorous contribution to the field of generalizable robot manipulation.
Research Focus
Key Achievements
Top Papers
- 1