Papers
4
Total Citations
113
H-Index
4
About
Haoran Song is a roboticist whose research lies at the intersection of motion planning, manipulation, and multi-agent coordination. His work is defined by tackling complex, real-world robotic challenges—from sorting cluttered objects to guiding autonomous swarms. Song’s most influential contribution is his pioneering use of Monte Carlo Tree Search (MCTS) for multi-object rearrangement, specifically in planar non-prehensile sorting. This work, which has garnered over 60 combined citations, demonstrates how a robot can intelligently push densely packed objects into separated classes without grasping them, a critical skill for logistics and manufacturing. He has also made significant strides in whole-arm manipulation, using reinforcement learning to enable robots to move large, bulky objects like a human body by leveraging the entire arm. Additionally, his "herding by caging" framework (26 citations) introduces a formation-based planning approach for guiding mobile agents, offering a novel solution for swarm robotics and environmental monitoring. Through these contributions, Song is advancing the frontier of robotic dexterity and autonomy, providing scalable solutions for tasks that require both precision and adaptability.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4